Split (1)

main · 5.79k rows

type stringclasses 1 value	name stringlengths 14 183	virtualsite_url stringlengths 46 46	speakers/authors stringlengths 8 1.31k	abstract stringlengths 246 3.59k
Poster	BIPNN: Learning to Solve Binary Integer Programming via Hypergraph Neural Networks	https://neurips.cc//virtual/2025/poster/116206	Sen Bai, Chunqi Yang, Xin Bai, Xin Zhang, Zhengang Jiang	Binary (0-1) integer programming (BIP) is pivotal in scientific domains requiring discrete decision-making. As the advance of AI computing, recent works explore neural network-based solver for integer linear programming (ILP) problems. Yet, they lack scalability for tackling nonlinear challenges. To handle nonlinearities, state-of-the-art Branch-and-Cut solvers employ linear relaxations, leading to exponential growth in auxiliary variables and severe computation limitations. To overcome these limitations, we propose BIPNN, an unsupervised learning framework to solve BIP problems via hypergraph neural networks (HyperGNN). Specifically, (i) BIPNN reformulates BIPs-constrained, discrete, and nonlinear (sin, log, exp) optimization problems-into unconstrained, differentiable, and polynomial loss functions. The reformulation stems from the observation of a precise one-to-one mapping between polynomial BIP objectives and hypergraph structures, enabling the unsupervised training of HyperGNN to optimize BIP problems in an end-to-end manner. On this basis, (ii) we propose a GPU-accelerated and continuous-annealing-enhanced training pipeline for BIPNN. The pipeline enables BIPNN to optimize large-scale nonlinear terms in BIPs fully in parallel via straightforward gradient descent, thus significantly reducing the training cost while ensuring the generation of discrete, high-quality solutions. Extensive experiments on synthetic and real-world datasets highlight the superiority of our approach.
Poster	Bipolar Self-attention for Spiking Transformers	https://neurips.cc//virtual/2025/poster/116154	Shuai Wang, Malu Zhang, Jingya Wang, Dehao Zhang, Yimeng Shan, Jieyuan Zhang, Yichen Xiao, Honglin Cao, Haonan Zhang, Zeyu Ma, Yang Yang, Haizhou Li	Harnessing the event-driven computation paradigm, Spiking Neural Networks present a promising avenue toward energy-efficient Transformer architectures. However, existing Spiking Transformers still suffer significant performance gaps compared to their artificial neural network counterparts. Through comprehensive analysis, we attribute this gap to these two factors. First, the binary nature of spike trains limits Spiking Self-attention (SSA)’s capacity to capture negative–negative and positive–negative membrane potential interactions on Querys and Keys. Second, SSA typically omits Softmax functions to avoid energy-intensive multiply-accumulate operations, thereby failing to maintain row-stochasticity constraints on attention scores.To address these issues, we propose a Bipolar Self-Attention (BSA) paradigm, effectively modeling multi-polar membrane potential interactions with a fully spike-driven characteristic. Specifically, we demonstrate that ternary matrix multiplication provides a closer approximation to real-valued computation on both distribution and local correlation, enabling clear differentiation between homopolar and heteropolar interactions. Moreover, we propose a shift-based Softmax approximation named Shiftmax, which efficiently achieves low-entropy activation and partly maintains row-stochasticity without non-linear operation, enabling precise attention allocation. Extensive experiments show that BSA achieves substantial performance improvements across various tasks, including image classification, semantic segmentation, and event-based tracking. These results establish its potential as a fundamental building block for energy-efficient Spiking Transformers.
Poster	Bisecle: Binding and Separation in Continual Learning for Video Language Understanding	https://neurips.cc//virtual/2025/poster/116080	Yue Tan, Xiaoqian Hu, Hao Xue, Celso de Melo, Flora Salim	Frontier vision-language models (VLMs) have made remarkable improvements in video understanding tasks. However, real-world videos typically exist as continuously evolving data streams (e.g., dynamic scenes captured by wearable glasses), necessitating models to continually adapt to shifting data distributions and novel scenarios. Considering the prohibitive computational costs of fine-tuning models on new tasks, usually, a small subset of parameters is updated while the bulk of the model remains frozen. This poses new challenges to existing continual learning frameworks in the context of large multimodal foundation models, i.e., catastrophic forgetting and update conflict. While the foundation models struggle with parameter-efficient continual learning, the hippocampus in the human brain has evolved highly efficient mechanisms for memory formation and consolidation. Inspired by the rapid Binding and pattern separation mechanisms in the hippocampus, in this work, we propose Bisecle for video-language continual learning, where a multi-directional supervision module is used to capture more cross-modal relationships and a contrastive prompt learning scheme is designed to isolate task-specific knowledge to facilitate efficient memory storage. Binding and separation processes further strengthen the ability of VLMs to retain complex experiences, enabling robust and efficient continual learning in video understanding tasks. We perform a thorough evaluation of the proposed Bisecle, demonstrating its ability to mitigate forgetting and enhance cross-task generalization on several VideoQA benchmarks.
Poster	BitMark for Infinity: Watermarking Bitwise Autoregressive Image Generative Models	https://neurips.cc//virtual/2025/poster/117685	Louis Kerner, Michel Meintz, Bihe Zhao, Franziska Boenisch, Adam Dziedzic	State-of-the-art text-to-image models like Infinity generate photorealistic images at an unprecedented speed. These models operate in a bitwise autoregressive manner over a discrete set of tokens that is practically infinite in size. However, their impressive generative power comes with a growing risk: as their outputs increasingly populate the Internet, they are likely to be scraped and reused as training data-potentially by the very same models. This phenomenon has been shown to lead to model collapse, where repeated training on generated content, especially from the models' own previous versions, causes a gradual degradation in performance. A promising mitigation strategy is watermarking, which embeds human-imperceptible yet detectable signals into generated images-enabling the identification of generated content. In this work, we introduce BitMark, a robust bitwise watermarking framework for Infinity. Our method embeds a watermark directly at the bit level of the token stream across multiple scales (also referred to as resolutions) during Infinity's image generation process. Our bitwise watermark subtly influences the bits to preserve visual fidelity and generation speed while remaining robust against a spectrum of removal techniques. Furthermore, it exhibits high radioactivity, i.e., when watermarked generated images are used to train another image generative model, this second model's outputs will also carry the watermark. The radioactive traces remain detectable even when only fine-tuning diffusion or image autoregressive models on images watermarked with our BitMark. Overall, our approach provides a principled step toward preventing model collapse in image generative models by enabling reliable detection of generated outputs.
Poster	Bits Leaked per Query: Information-Theoretic Bounds for Adversarial Attacks on LLMs	https://neurips.cc//virtual/2025/poster/119179	Masahiro Kaneko, Timothy Baldwin	Adversarial attacks by malicious users that threaten the safety of large language models (LLMs) can be viewed as attempts to infer a \emph{target property} $T$ that is unknown when an instruction is issued, and becomes knowable only after the model's reply is observed. Examples of target properties $T$ include the binary flag that triggers an LLM's harmful response or rejection, and the degree to which information deleted by unlearning can be restored, both elicited via adversarial instructions. The LLM reveals an \emph{observable signal} $Z$ that potentially leaks hints for attacking through a response containing answer tokens, thinking process tokens, or logits.Yet the scale of information leaked remains anecdotal, leaving auditors without principled guidance and defenders blind to the transparency--risk trade-off.We fill this gap with an information-theoretic framework\footnote{We plan to release the code submitted as supplementary material after our paper is accepted.} that computes how much information can be safely disclosed, and enables auditors to gauge how close their methods come to the fundamental limit.Treating the mutual information $I(Z;T)$ between the observation $Z$ and the target property $T$ as the leaked bits per query, we show that achieving error $\varepsilon$ requires at least $\log(1/\varepsilon)/I(Z;T)$ queries, scaling linearly with the inverse leak rate and only logarithmically with the desired accuracy.Thus, even a modest increase in disclosure collapses the attack cost from quadratic to logarithmic in terms of the desired accuracy.Experiments on seven LLMs across system-prompt leakage, jailbreak, and relearning attacks corroborate the theory: exposing answer tokens alone requires about a thousand queries, adding logits cuts this to about a hundred, and revealing the full thinking process trims it to a few dozen.Our results provide the first principled yardstick for balancing transparency and security when deploying LLMs.
Poster	Bit-swapping Oriented Twin-memory Multi-view Clustering in Lifelong Incomplete Scenarios	https://neurips.cc//virtual/2025/poster/119818	Shengju Yu, Pei Zhang, Siwei Wang, Suyuan Liu, Xinhang Wan, Zhibin Dong, Tiejun Li, Xinwang Liu	Although receiving notable improvements, current multi-view clustering (MVC) techniques generally rely on feature library mechanisms to propagate accumulated knowledge from historical views to newly-arrived data, which overlooks the information pertaining to basis embedding within each view. Moreover, the mapping paradigm inevitably alters the values of learned landmarks and built affinities due to the uninterruption nature, accordingly disarraying the hierarchical cluster structures. To mitigate these two issues, we in the paper provide a named BSTM algorithm. Concretely, we firstly synchronize with the distinct dimensions by introducing a group of specialized projectors, and then establish unified anchors for all views collected so far to capture intrinsic patterns. Afterwards, departing from per-view architectures, we devise a shared bipartite graph construction via indicators to quantify similarity, which not only avoids redundant data-recalculations but alleviates the representation distortion caused by fusion. Crucially, there two components are optimized within an integrated framework, and collectively facilitate knowledge transfer upon encountering incoming views. Subsequently, to flexibly do transformation on anchors and meanwhile maintain numerical consistency, we develop a bit-swapping scheme operating exclusively on 0 and 1. It harmonizes anchors on current view and that on previous views through one-hot encoded row and column attributes, and the graph structures are correspondingly reordered to reach a matched configuration. Furthermore, a computationally efficient four-step updating strategy with linear complexity is designed to minimize the associated loss. Extensive experiments organized on publicly-available benchmark datasets with varying missing percentages confirm the superior effectiveness of our BSTM.
Poster	Bivariate Matrix-valued Linear Regression (BMLR): Finite-sample performance under Identifiability and Sparsity Assumptions	https://neurips.cc//virtual/2025/poster/120128	Nayel Bettache	This paper studies a bilinear matrix-valued regression model where both predictors and responses are matrix-valued. For each observation $ t = 1, \ldots, T $, the response $ Y_t \in \mathbb{R}^{n \times p} $ and predictor $ X_t \in \mathbb{R}^{m \times q} $ satisfy the model $Y_t = A^* X_t B^* + E_t,$with $ A^* \in \mathbb{R}_+^{n \times m} $ (row-wise $\ell_1$-normalized), $ B^* \in \mathbb{R}^{q \times p} $, and $ E_t $ independent Gaussian noise matrices. The goal is to estimate $ A^* $ and $ B^* $ from the observed pairs $ (X_t, Y_t) $.We propose explicit, optimization-free estimators and establish non-asymptotic error bounds, including sparse settings. Simulations confirm the theoretical rates and demonstrate strong finite-sample performance. We further illustrate the practical utility of our method through an image denoising application on real data.
Poster	Black-Box Membership Inference Attack for LVLMs via Prior Knowledge-Calibrated Memory Probing	https://neurips.cc//virtual/2025/poster/119960	Jinhua Yin, Peiru Yang, Chen Yang, Huili Wang, Zhiyang Hu, Shangguang Wang, Yongfeng Huang, Tao Qi	Large vision-language models (LVLMs) derive their capabilities from extensive training on vast corpora of visual and textual data. Empowered by large-scale parameters, these models often exhibit strong memorization of their training data, rendering them susceptible to membership inference attacks (MIAs).Existing MIA methods for LVLMs typically operate under white- or gray-box assumptions, by extracting likelihood-based features for the suspected data samples based on the target LVLMs. However, mainstream LVLMs generally only expose generated outputs while concealing internal computational features during inference, limiting the applicability of these methods.In this work, we propose the first black-box MIA framework for LVLMs, based on a prior knowledge-calibrated memory probing mechanism. The core idea is to assess the model memorization of the private semantic information embedded within the suspected image data, that is unlikely to be inferred from general world knowledge alone.We conduct extensive experiments across four LVLMs and three datasets.Empirical results demonstrate that our method effectively identifies training data of LVLMs in a purely black-box setting and even achieves performance comparable to gray-box and white-box methods.Further analysis reveals the robustness of our method against potential adversarial manipulations, and the effectiveness of the methodology designs.Our code and data are available at \url{https://anonymous.4open.science/r/KCMP-2D2C}
Poster	Blackbox Model Provenance via Palimpsestic Membership Inference	https://neurips.cc//virtual/2025/poster/117688	Rohith Kuditipudi, Jing Huang, Sally Zhu, Percy Liang, Christopher Potts, Diyi Yang	Suppose Alice trains an open-weight language model, and subsequently Bob uses a blackbox derivative of Alice's model to produce text. Can Alice prove that Bob is using her model, either by querying Bob's derivative model (query setting) or from the text alone (sample setting)? We formulate this question as an independence testing problem---in which the null hypothesis is that Bob's model is independent of Alice's randomized training run---and investigate it through the lens of \textit{palimpsestic memorization} in language models: models are more likely to memorize data seen later in training, so we can test whether Bob's model derives from Alice's using test statistics that capture correlation between the output of Bob's model and the ordering of examples in Alice's training run. So long as Alice has randomly shuffled her training data, any significant correlation amounts to exactly quantifiable statistical evidence against the null hypothesis, regardless of the composition of Alice's training data. We develop tests for both the query and sample settings and empirically validate the power of our tests using the Pythia and OLMo model families, as well as small-scale models trained on TinyStories. In the query setting, we query Bob's model on Alice's training data and measure the correlation of its log-likelihood with the ordering of data. We show that this test is robust to common post-training practices (e.g., supervised fine-tuning, preference optimization, model souping). In the sample setting, we match spans of Bob's text against Alice's training examples and correlate the likelihood of a match with the ordering of training examples. We show this test reliably attributes text to models given a few thousand tokens. Our work offers a novel framework for provenance verification of open-weight language models, enabling accountability and protection for models.
Poster	Blameless Users in a Clean Room: Defining Copyright Protection for Generative Models	https://neurips.cc//virtual/2025/poster/117712	Aloni Cohen	Are there any conditions under which a generative model’s outputs are guaranteed not to infringe the copyrights of its training data? This is the question of "provable copyright protection" first posed by Vyas, Kakade, and Barak [ICML 2023]. They define _near access-freeness (NAF)_ and propose it as sufficient for protection. This paper revisits the question and establishes new foundations for provable copyright protection---foundations that are firmer both technically and legally. First, we show that NAF alone does not prevent infringement. In fact, NAF models can enable verbatim copying, a blatant failure of copy protection that we dub being _tainted_. Then, we introduce our _blameless copy protection framework_ for defining meaningful guarantees, and instantiate it with _clean-room copy protection_. Clean-room copy protection allows a user to control their risk of copying by behaving in a way that is unlikely to copy in a counterfactual "clean-room setting." Finally, we formalize a common intuition about differential privacy and copyright by proving that DP implies clean-room copy protection when the dataset is _golden_, a copyright deduplication requirement.
Poster	Blending Complementary Memory Systems in Hybrid Quadratic-Linear Transformers	https://neurips.cc//virtual/2025/poster/118163	Kazuki Irie, Morris Yau, Samuel J Gershman	We develop hybrid memory architectures for general-purpose sequence processing neural networks, that combine key-value memory using softmax attention (KV-memory) with dynamic synaptic memory through fast-weight programming (FW-memory)---the core principles of quadratic and linear transformers, respectively. These two memory systems have complementary but individually limited properties: KV-memory offers precise retrieval but is constrained by quadratic complexity in sequence length, while FW-memory supports arbitrarily long sequences and enables more expressive computation but sacrifices precise recall. We propose and compare three methods to blend these two systems into a single memory system to leverage the strengths of both. We conduct experiments on general language modeling and retrieval tasks by training 340M- and 1.3B-parameter models from scratch, as well as on synthetic algorithmic tasks designed to precisely illustrate the benefits of certain hybrid methods over others. Overall, we demonstrate how a well-designed hybrid can overcome the limitations of its individual components, offering new insights into the design principle of neural memory systems.
Poster	BLEUBERI: BLEU is a surprisingly effective reward for instruction following	https://neurips.cc//virtual/2025/poster/117192	Yapei Chang, Yekyung Kim, Michael Krumdick, Amir Zadeh, Chuan Li, Chris Tanner, Mohit Iyyer	Reward models are central to aligning LLMs with human preferences, but they are costly to train, requiring large-scale human-labeled preference data and powerful pretrained LLM backbones. Meanwhile, the increasing availability of high-quality synthetic instruction-following datasets raises the question: can simpler, reference-based metrics serve as viable alternatives to reward models during RL-based alignment? In this paper, we show first that BLEU, a basic string-matching metric, surprisingly matches strong reward models in agreement with human preferences on general instruction-following datasets. Based on this insight, we develop BLEUBERI, a method that first identifies challenging instructions and then applies Group Relative Policy Optimization (GRPO) using BLEU directly as the reward function. We demonstrate that BLEUBERI-trained models are competitive with models trained via reward model-guided RL across four challenging instruction-following benchmarks and three different base language models. A human evaluation further supports that the quality of BLEUBERI model outputs is on par with those from reward model-aligned models. Moreover, BLEUBERI models generate outputs that are more factually grounded than competing methods. Overall, we show that given access to high-quality reference outputs (easily obtained via existing instruction-following datasets or synthetic data generation), string matching-based metrics are cheap yet effective proxies for reward models during alignment.
Poster	Blindfolded Experts Generalize Better: Insights from Robotic Manipulation and Videogames	https://neurips.cc//virtual/2025/poster/119957	Ev Zisselman, Mirco Mutti, Shelly Francis-Meretzki, Elisei Shafer, Aviv Tamar	Behavioral cloning is a simple yet effective technique for learning sequential decision-making from demonstrations. Recently, it has gained prominence as the core of foundation models for the physical world, where achieving generalization requires countless demonstrations of a multitude of tasks. Typically, a human expert with full information on the task demonstrates a (nearly) optimal behavior. In this paper, we propose to hide some of the task's information from the demonstrator. This ``blindfolded'' expert is compelled to employ non-trivial exploration to solve the task. We show that cloning the blindfolded expert generalizes better to unseen tasks than its fully-informed counterpart. We conduct experiments of real-world robot peg insertion tasks with (limited) human demonstrations, alongside a videogame from the Procgen benchmark. Additionally, we support our findings with theoretical analysis, which confirms that the generalization error scales with $\sqrt{I/m}$, where $I$ measures the amount of task information available to the demonstrator, and $m$ is the number of demonstrated tasks. Both theory and practice indicate that cloning blindfolded experts generalizes better with fewer demonstrated tasks. Project page with videos and code: [https://sites.google.com/view/blindfoldedexperts/home](https://sites.google.com/view/blindfoldedexperts/home)
Poster	BLINK-Twice: You see, but do you observe? A Reasoning Benchmark on Visual Perception	https://neurips.cc//virtual/2025/poster/121522	junyan ye, DONGZHI JIANG, Jun He, Baichuan Zhou, Zilong Huang, Zhiyuan Yan, Hongsheng Li, Conghui He, Weijia Li	Recently, Multimodal Large Language Models (MLLMs) have made rapid progress, particularly in enhancing their reasoning capabilities. However, existing reasoning benchmarks still primarily assess language-based reasoning, often treating visual input as replaceable context. To address this gap, we introduce BLINK-Twice, a vision-centric reasoning benchmark grounded in challenging perceptual tasks. Instead of relying on external knowledge, our tasks require models to reason from visual content alone, shifting the focus from language-based to image-grounded reasoning. Compared to prior perception benchmarks, it moves beyond shallow perception ("see") and requires fine-grained observation and analytical reasoning ("observe"). BLINK-Twice integrates three core components: seven types of visual challenges for testing visual reasoning, natural adversarial image pairs that enforce reliance on visual content, and annotated reasoning chains for fine-grained evaluation of the reasoning process rather than final answers alone. We evaluate 20 leading MLLMs, including 12 foundation models and 8 reasoning-enhanced models. BLINK-Twice poses a significant challenge to current models. While existing reasoning strategies in the language space—such as chain-of-thought or self-criticism can improve performance, they often result in unstable and redundant reasoning. We observe that repeated image observation improves performance across models, and active visual interaction, as demonstrated by models like o3, highlights the need for a new paradigm for vision reasoning. The dataset is publicly available at https://huggingface.co/datasets/PicoTrex/BLINK-Twice.
Poster	Block-Biased Mamba for Long-Range Sequence Processing	https://neurips.cc//virtual/2025/poster/119862	Annan Yu, N. Benjamin Erichson	Mamba extends earlier state space models (SSMs) by introducing input-dependent dynamics, and has demonstrated strong empirical performance across a range of domains, including language modeling, computer vision, and foundation models. However, a surprising weakness remains: despite being built on architectures designed for long-range dependencies, Mamba performs poorly on long-range sequential tasks. Understanding and addressing this gap is important for improving Mamba's universality and versatility. In this work, we analyze Mamba’s limitations through three perspectives: expressiveness, inductive bias, and training stability. Our theoretical results show how Mamba falls short in each of these aspects compared to earlier SSMs such as S4D. To address these issues, we propose $\text{B}\_{2}\text{S}\_{6}$, a simple extension of Mamba's S6 unit that combines block-wise selective dynamics with a channel-specific bias. We prove that these changes equip the model with a better-suited inductive bias and improve its expressiveness and stability. Empirically, $\text{B}\_{2}\text{S}\_{6}$ outperforms S4 and S4D on Long-Range Arena (LRA) tasks while maintaining Mamba's performance on language modeling benchmarks.
Poster	Block Coordinate Descent for Neural Networks Provably Finds Global Minima	https://neurips.cc//virtual/2025/poster/115652	Shunta Akiyama	In this paper, we consider a block coordinate descent (BCD) algorithm for training deep neural networks and provide a new global convergence guarantee under strictly monotonically increasing activation functions. While existing works demonstrate convergence to stationary points for BCD in neural networks, our contribution is the first to prove convergence to global minima, ensuring arbitrarily small loss. We show that the loss with respect to the output layer decreases exponentially while the loss with respect to the hidden layers remains well-controlled. Additionally, we derive generalization bounds using the Rademacher complexity framework, demonstrating that BCD not only achieves strong optimization guarantees but also provides favorable generalization performance. Moreover, we propose a modified BCD algorithm with skip connections and non-negative projection, extending our convergence guarantees to ReLU activation, which are not strictly monotonic. Empirical experiments confirm our theoretical findings, showing that the BCD algorithm achieves a small loss for strictly monotonic and ReLU activations.
Poster	BlockDecoder: Boosting ASR Decoders with Context and Merger Modules	https://neurips.cc//virtual/2025/poster/117110	Darshan Prabhu, Preethi Jyothi	Attention-based encoder decoder models are the backbone of state-of-the-art architectures for automatic speech recognition (ASR). These models combine a powerful encoder that extracts rich acoustic features with a decoder that autoregressively produces the ASR output. The decoder handles two critical tasks: (1) building rich text-only context and (2) merging acoustic information from the encoder to ensure the predictions remain faithful to the audio. We observe a systematic pattern across the attention distributions of decoder layers in prior architectures: the initial layers direct most attention towards building textual context, while the latter layers focus largely on merging acoustic and textual information to make accurate predictions. Leveraging this key insight, we propose BlockDecoder, a novel decoder architecture comprising two distinct components: a TextEncoder that is purely text-based, and a Merger that autoregressively generates tokens in blockwise fashion by combining information from the audio encoder and TextEncoder. These two components of BlockDecoder collectively result in substantial latency gains. Across diverse datasets, languages and speech tasks, we demonstrate that our proposed BlockDecoder achieves a significant speedup ($\sim 2$x) compared to traditional decoders, without any degradation in performance.
Poster	Block-Diagonal LoRA for Eliminating Communication Overhead in Tensor Parallel LoRA Serving	https://neurips.cc//virtual/2025/poster/120211	Xinyu Wang, Jonas M. Kübler, Kailash Budhathoki, Yida Wang, Matthäus Kleindessner	When serving a single base LLM with several different LoRA adapters simultaneously, the adapters cannot simply be merged with the base model’s weights as the adapter swapping would create overhead and requests using different adapters could not be batched. Rather, the LoRA computations have to be separated from the base LLM computations, and in a multi-device setup the LoRA adapters can be sharded in a way that is well aligned with the base model’s tensor parallel execution, as proposed in S-LoRA. However, the S-LoRA sharding strategy encounters some communication overhead, which may be small in theory, but can be large in practice. In this paper, we propose to constrain certain LoRA factors to be block-diagonal, which allows for an alternative way of sharding LoRA adapters that does not require any additional communication for the LoRA computations. We demonstrate in extensive experiments that our block-diagonal LoRA approach is similarly parameter efficient as standard LoRA (i.e., for a similar number of parameters it achieves similar downstream performance) and that it leads to significant end-to-end speed-up over S-LoRA. For example, when serving on eight A100 GPUs, we observe up to 1.79x (1.23x) end-to-end speed-up with 0.87x (1.74x) number of adapter parameters for Llama-3.1-70B, and up to 1.63x (1.3x) end-to-end speed-up with 0.86x (1.73x) number of adapter parameters for Llama-3.1-8B.
Poster	BlockScan: Detecting Anomalies in Blockchain Transactions	https://neurips.cc//virtual/2025/poster/117763	Jiahao Yu, Xian Wu, Hao Liu, Wenbo Guo, Xinyu Xing	We propose BlockScan, a customized Transformer for anomaly detection in blockchain transactions.Unlike existing methods that rely on rule-based systems or directly apply off-the-shelf large language models (LLMs), BlockScan introduces a series of customized designs to effectively model the unique data structure of blockchain transactions.First, a blockchain transaction is multi-modal, containing blockchain-specific tokens, texts, and numbers.We design a novel modularized tokenizer to handle these multi-modal inputs, balancing the information across different modalities.Second, we design a customized masked language modeling mechanism for pretraining the Transformer architecture, incorporating RoPE embedding and FlashAttention for handling longer sequences.Finally, we design a novel anomaly detection method based on the model outputs.We further provide theoretical analysis for the detection method of our system. Extensive evaluations on Ethereum and Solana transactions demonstrate BlockScan's exceptional capability in anomaly detection while maintaining a low false positive rate.Remarkably, BlockScan is the only method that successfully detects anomalous transactions on Solana with high accuracy, whereas all other approaches achieved very low or zero detection recall scores.This work sets a new benchmark for applying Transformer-based approaches in blockchain data analysis.
Poster	Blockwise Flow Matching: Improving Flow Matching Models For Efficient High-Quality Generation	https://neurips.cc//virtual/2025/poster/118395	Dogyun Park, Taehoon Lee, Minseok Joo, Hyunwoo Kim	Recently, Flow Matching models have pushed the boundaries of high-fidelity data generation across a wide range of domains. It typically employs a single large network to learn the entire generative trajectory from noise to data. Despite their effectiveness, this design struggles to capture distinct signal characteristics across timesteps simultaneously and incurs substantial inference costs due to the iterative evaluation of the entire model. To address these limitations, we propose Blockwise Flow Matching (BFM), a novel framework that partitions the generative trajectory into multiple temporal segments, each modeled by smaller but specialized velocity blocks. This blockwise design enables each block to specialize effectively in its designated interval, improving inference efficiency and sample quality. To further enhance generation fidelity, we introduce a Semantic Feature Guidance module that explicitly conditions velocity blocks on semantically rich features aligned with pretrained representations.Additionally, we propose a lightweight Feature Residual Approximation strategy that preserves semantic quality while significantly reducing inference cost.Extensive experiments on ImageNet 256$\times$256 demonstrate that BFM establishes a substantially improved Pareto frontier over existing Flow Matching methods, achieving 2.1$\times$ to 4.9$\times$ accelerations in inference complexity at comparable generation performance.
Poster	BlurDM: A Blur Diffusion Model for Image Deblurring	https://neurips.cc//virtual/2025/poster/120268	Jin-Ting He, Fu-Jen Tsai, Yan-Tsung Peng, Min-Hung Chen, Chia-Wen Lin, Yen-Yu Lin	Diffusion models show promise for dynamic scene deblurring; however, existing studies often fail to leverage the intrinsic nature of the blurring process within diffusion models, limiting their full potential. To address it, we present a Blur Diffusion Model (BlurDM), which seamlessly integrates the blur formation process into diffusion for image deblurring. Observing that motion blur stems from continuous exposure, BlurDM implicitly models the blur formation process through a dual-diffusion forward scheme, diffusing both noise and blur onto a sharp image. During the reverse generation process, we derive a dual denoising and deblurring formulation, enabling BlurDM to recover the sharp image by simultaneously denoising and deblurring, given pure Gaussian noise conditioned on the blurred image as input. Additionally, to efficiently integrate BlurDM into deblurring networks, we perform BlurDM in the latent space, forming a flexible prior generation network for deblurring. Extensive experiments demonstrate that BlurDM significantly and consistently enhances existing deblurring methods on four benchmark datasets.
Poster	BlurGuard: A Simple Approach for Robustifying Image Protection Against AI-Powered Editing	https://neurips.cc//virtual/2025/poster/115381	Jinsu Kim, Yunhun Nam, Minseon Kim, Sangpil Kim, Jongheon Jeong	Recent advances in text-to-image models have increased the exposure of powerful image editing techniques as a tool, raising concerns about their potential for malicious use. An emerging line of research to address such threats focuses on implanting “protective” adversarial noise into images before their public release, so future attempts to edit them using text-to-image models can be impeded. However, subsequent works have shown that these adversarial noises are often easily “reversed,” e.g., with techniques as simple as JPEG compression, casting doubt on the practicality of the approach. In this paper, we argue that adversarial noise for image protection should not only be imperceptible, as has been a primary focus of prior work, but also irreversible, viz., it should be difficult to detect as noise provided that the original image is hidden. We propose a surprisingly simple method to enhance the robustness of image protection methods against noise reversal techniques. Specifically, it applies an adaptive per-region Gaussian blur on the noise to adjust the overall frequency spectrum. Through extensive experiments, we show that our method significantly improves the per-sample worst-case protection performance of existing methods against a wide range of reversal techniques, while also reducing quality degradation due to noise in terms of perceptual metrics.
Poster	BMMR: A Large-Scale Bilingual Multimodal Multi-Discipline Reasoning Dataset	https://neurips.cc//virtual/2025/poster/121856	Zhiheng Xi, Guanyu Li, Yutao Fan, Honglin Guo, Yufang Liu, Xiaoran Fan, Jiaqi Liu, dingjinchao, Wangmeng Zuo, Zhenfei Yin, LEI BAI, Tao Ji, Tao Gui, Qi Zhang, Xuanjing Huang	In this paper, we introduce BMMR, a large-scale bilingual, multimodal, multi-disciplinary reasoning dataset for the community to develop and evaluate large multimodal models (LMMs). BMMR comprises 100k university-level questions drawn from 300 UNESCO-defined subjects, spanning diverse formats—multiple-choice, fill-in-the-blank, and open-ended QA—and sourced from both print and digital media such as books, exams, and quizzes. All data are curated and filtered via a human-in-the-loop, automated, and scalable framework, and each instance is paired with a high-quality reasoning path. The dataset is organized into two parts: BMMR-Eval that comprises 20k high-quality instances to comprehensively assess LMMs’ knowledge and reasoning across multiple disciplines in both Chinese and English; and BMMR-Train that contains 80k instances to support further research and development, extending the current focus on mathematical reasoning to diverse disciplines and domains. In addition, we propose the process-based multi-discipline BMMR-Verifier for accurate and fine-grained evaluation of LMMs’ reasoning. Extensive experiments reveal that (i) even SOTA models leave substantial headroom on BMMR-Eval; (ii) reasoning models exhibit discipline bias and outperform LMMs only on specific subjects; (iii) open-source models still trail their proprietary counterparts; and (iv) fine-tuning on BMMR-Train narrows this gap. Additionally, we conduct reasoning-chain analyses using BMMR-Verifier and other in-depth studies, uncovering the challenges LMMs currently face in multidisciplinary reasoning. We will release the data and models, and we believe our work can offers valuable insights and contributions to the community.
Poster	BMW: Bidirectionally Memory bank reWriting for Unsupervised Person Re-Identification	https://neurips.cc//virtual/2025/poster/119766	Xiaobin Liu, Jianing Li, Baiwei Guo, WenbinZhu, Jing Yuan	Recent works show that contrastive learning based on memory banks is an effective framework for unsupervised person Re-IDentification (ReID). In existing methods, memory banks are typically initialized with cluster centroids and rewritten with positive samples via the momentum mechanism along with the model training. However, this mechanism solely focuses on the intra-class compactness by pulling memory banks close to positive samples, neglecting the inter-class separability among different memory banks. Rewriting memory banks with partial constraint limits their discrimination capacities, and hence hinders learning discriminative features based on those memory banks. In this paper, we claim that memory banks should be rewritten with both intra-class and inter-class constraints, and therefore propose a unified memory bank rewriting mechanism, Bidirectionally Memory bank reWriting (BMW), to chase enhanced discrimination capacity. Specifically, BMW formulates the memory bank rewriting as the gradient descent update with two objectives, i.e., reducing intra-class diversity and enhancing inter-class separability. To effectively enhance the separability of memory banks with limited number of rewriting steps, we further design a novel objective formulation for the inter-class constraint, which is more effective for one step update. BMW enhances both representation and discrimination capacities of memory banks, thus leads to an effective ReID feature optimization. BMW is simple yet effective and can serve as a new paradigm for person ReID methods based on memory banks. Extensive experiments on standard benchmarks demonstrate the effectiveness of our BMW method in unsupervised ReID model training. Specially, BMW even outperforms previous methods that use stronger backbones. Code will be available.
Poster	BNMusic: Blending Environmental Noises into Personalized Music	https://neurips.cc//virtual/2025/poster/118824	Chi Zuo, Martin Møller, Pablo Martínez-Nuevo, Huayang Huang, Yu Wu, Ye Zhu	While being disturbed by environmental noises, the acoustic masking technique is a conventional way to reduce the annoyance in audio engineering that seeks to cover up the noises with other dominant yet less intrusive sounds. However, misalignment between the dominant sound and the noise—such as mismatched downbeats—often requires an excessive volume increase to achieve effective masking. Motivated by recent advances in cross-modal generation, in this work, we introduce an alternative method to acoustic masking, aiming to reduce the noticeability of environmental noises by blending them into personalized music generated based on user-provided text prompts. Following the paradigm of music generation using mel-spectrogram representations, we propose a Blending Noises into Personalized Music (BNMusic) framework with two key stages. The first stage synthesizes a complete piece of music in a mel-spectrogram representation that encapsulates the musical essence of the noise. In the second stage, we adaptively amplifying the generated music segment to further reduce noise perception and enhance the blending effectiveness, while preserving auditory quality. Our experiments with comprehensive evaluations on MusicBench, EPIC-SOUNDS, and ESC-50 demonstrate the effectiveness of our framework, highlighting the ability to blend environmental noise with rhythmically aligned, adaptively amplified, and enjoyable music segments, minimizing the noticeability of the noise, thereby improving overall acoustic experiences.
Poster	BO4Mob: Bayesian Optimization Benchmarks for High-Dimensional Urban Mobility Problem	https://neurips.cc//virtual/2025/poster/121585	Seunghee Ryu, Donghoon Kwon, Seongjin Choi, Aryan Deshwal, Seungmo Kang, Carolina Osorio	We introduce BO4Mob, a new benchmark framework for high-dimensional Bayesian Optimization (BO), driven by the challenge of origin-destination (OD) travel demand estimation in large urban road networks. Estimating OD travel demand from limited traffic sensor data is a difficult inverse optimization problem, particularly in real-world, large-scale transportation networks. This problem involves optimizing over high-dimensional continuous spaces where each objective evaluation is computationally expensive, stochastic, and non-differentiable. BO4Mob comprises five scenarios based on real-world San Jose, CA road networks, with input dimensions scaling up to 10,100. These scenarios utilize high-resolution, open-source traffic simulations that incorporate realistic nonlinear and stochastic dynamics. We demonstrate the benchmark's utility by evaluating four optimization methods: three state-of-the-art BO algorithms and one non-BO baseline. This benchmark is designed to support both the development of scalable optimization algorithms and their application for the design of data-driven urban mobility models, including high-resolution digital twins of metropolitan road networks. Code and documentation are available at https://github.com/UMN-Choi-Lab/BO4Mob
Poster	Bohdi: Heterogeneous LLM Fusion with Automatic Data Exploration	https://neurips.cc//virtual/2025/poster/115321	Junqi Gao, Zhichang Guo, Dazhi Zhang, Dong Li, Runze Liu, Pengfei Li, Kai Tian, Biqing Qi	Heterogeneous Large Language Model (LLM) fusion integrates the strengths of multiple source LLMs with different architectures into a target LLM with low computational overhead. While promising, existing methods suffer from two major limitations: 1) reliance on real data from limited domain for knowledge fusion, preventing the target LLM from fully acquiring knowledge across diverse domains, and 2) fixed data allocation proportions across domains, failing to dynamically adjust according to the target LLM's varying capabilities across domains, leading to a capability imbalance. To overcome these limitations, we propose Bohdi, a synthetic-data-only heterogeneous LLM fusion framework. Through the organization of knowledge domains into a hierarchical tree structure, Bohdi enables automatic domain exploration and multi-domain data generation through multi-model collaboration, thereby comprehensively extracting knowledge from source LLMs. By formalizing domain expansion and data sampling proportion allocation on the knowledge tree as a Hierarchical Multi-Armed Bandit problem, Bohdi leverages the designed DynaBranches mechanism to adaptively adjust sampling proportions based on the target LLM's performance feedback across domains. Integrated with our proposed Introspection-Rebirth (IR) mechanism, DynaBranches dynamically tracks capability shifts during target LLM's updates via Sliding Window Binomial Likelihood Ratio Testing (SWBLRT), further enhancing its online adaptation capability. Comparative experimental results on a comprehensive suite of benchmarks demonstrate that Bohdi significantly outperforms existing baselines on multiple target LLMs, exhibits higher data efficiency, and virtually eliminates the imbalance in the target LLM's capabilities.
Poster	BoltzNCE: Learning likelihoods for Boltzmann Generation with Stochastic Interpolants and Noise Contrastive Estimation	https://neurips.cc//virtual/2025/poster/119032	Rishal Aggarwal, Jacky Chen, Nicholas Boffi, David Koes	Efficient sampling from the Boltzmann distribution defined by an energy function is a key challenge in modeling physical systems such as molecules. Boltzmann Generators tackle this by leveraging Continuous Normalizing Flows that transform a simple prior into a distribution that can be reweighted to match the Boltzmann distribution using sample likelihoods. However, obtaining likelihoods requires computing costly Jacobians during integration, making it impractical for large molecular systems. To overcome this, we propose learning the likelihood of the generated distribution via an energy-based model trained with noise contrastive estimation and score matching. By using stochastic interpolants to anneal between the prior and generated distributions, we combine both the objective functions to efficiently learn the density function. On the alanine dipeptide system, we demonstrate that our method yields free energy profiles and energy distributions comparable to those obtained with exact likelihoods. Additionally, we show that free energy differences between metastable states can be estimated accurately with orders-of-magnitude speedup.
Poster	BOOM: Benchmarking Out-Of-distribution Molecular Property Predictions of Machine Learning Models	https://neurips.cc//virtual/2025/poster/121661	Evan Antoniuk, Shehtab Zaman, Tal Ben-Nun, Peggy Li, James Diffenderfer, Busra Sahin, Obadiah Smolenski, Everett Grethel, Tim Hsu, Anna Hiszpanski, Kenneth Chiu, Bhavya Kailkhura, Brian Van Essen	Data-driven molecular discovery leverages artificial intelligence/machine learning (AI/ML) and generative modeling to filter and design novel molecules. Discovering novel molecules requires accurate out-of-distribution (OOD) predictions, but ML models struggle to generalize OOD. Currently, no systematic benchmarks exist for molecular OOD prediction tasks. We present BOOM, $\textbf{b}$enchmarks for $\textbf{o}$ut-$\textbf{o}f$-$\textbf{d}$istribution $\textbf{m}$olecular property predictions: a chemically-informed benchmark for OOD performance on common molecular property prediction tasks. We evaluate over 140 model-task combinations to benchmark deep learning models on OOD performance. Overall, we find that no existing model achieves strong generalization across all tasks: even the top-performing model exhibited an average OOD error 3$\times$ higher than in-distribution. Current chemical foundation models do not show strong OOD extrapolation, while models with high inductive bias can perform well on OOD tasks with simple, specific properties. We perform extensive ablation experiments, highlighting how data generation, pre-training, hyperparameter optimization, model architecture, and molecular representation impact OOD performance. Developing models with strong OOD generalization is a new frontier challenge in chemical ML. This open-source benchmark is available at https://github.com/FLASK-LLNL/BOOM
Poster	Boosting Adversarial Transferability with Spatial Adversarial Alignment	https://neurips.cc//virtual/2025/poster/115672	Zhaoyu Chen, HaiJing Guo, Kaixun Jiang, Jiyuan Fu, Xinyu Zhou, Dingkang Yang, Hao Tang, Bo Li, Wenqiang Zhang	Deep neural networks are vulnerable to adversarial examples that exhibit transferability across various models. Numerous approaches are proposed to enhance the transferability of adversarial examples, including advanced optimization, data augmentation, and model modifications. However, these methods still show limited transferability, partiovovocularly in cross-architecture scenarios, such as from CNN to ViT. To achieve high transferability, we propose a technique termed Spatial Adversarial Alignment (SAA), which employs an alignment loss and leverages a witness model to fine-tune the surrogate model. Specifically, SAA consists of two key parts: spatial-aware alignment and adversarial-aware alignment. First, we minimize the divergences of features between the two models in both global and local regions, facilitating spatial alignment. Second, we introduce a self-adversarial strategy that leverages adversarial examples to impose further constraints, aligning features from an adversarial perspective. Through this alignment, the surrogate model is trained to concentrate on the common features extracted by the witness model. This facilitates adversarial attacks on these shared features, thereby yielding perturbations that exhibit enhanced transferability. Extensive experiments on various architectures on ImageNet show that aligned surrogate models based on SAA can provide higher transferable adversarial examples, especially in cross-architecture attacks. Our source code is available at Supplementary Materials.
Poster	Boosting Generative Image Modeling via Joint Image-Feature Synthesis	https://neurips.cc//virtual/2025/poster/116596	Theodoros Kouzelis, Efstathios Karypidis, Ioannis Kakogeorgiou, Spyridon Gidaris, Nikos Komodakis	Latent diffusion models (LDMs) dominate high-quality image generation, yet integrating representation learning with generative modeling remains a challenge. We introduce a novel generative image modeling framework that seamlessly bridges this gap by leveraging a diffusion model to jointly model low-level image latents (from a variational autoencoder) and high-level semantic features (from a pretrained self-supervised encoder like DINO). Our latent-semantic diffusion approach learns to generate coherent image-feature pairs from pure noise, significantly enhancing both generative quality and training efficiency, all while requiring only minimal modifications to standard Diffusion Transformer architectures. By eliminating the need for complex distillation objectives, our unified design simplifies training and unlocks a powerful new inference strategy: Representation Guidance, which leverages learned semantics to steer and refine image generation. Evaluated in both conditional and unconditional settings, our method delivers substantial improvements in image quality and training convergence speed, establishing a new direction for representation-aware generative modeling.
Poster	Boosting Knowledge Utilization in Multimodal Large Language Models via Adaptive Logits Fusion and Attention Reallocation	https://neurips.cc//virtual/2025/poster/115855	Wenbin An, Jiahao Nie, Feng Tian, Haonan Lin, mingxiang cai, Yaqiang Wu, QianYing Wang, Xiaoqin Zhang, Shijian Lu	Despite their recent progress, Multimodal Large Language Models (MLLMs) often struggle in knowledge-intensive tasks due to the limited and outdated parametric knowledge they acquired during training. Multimodal Retrieval Augmented Generation addresses this issue by retrieving contextual knowledge from external databases, thereby enhancing LVLMs with expanded knowledge sources. However, existing LVLMs often fail to fully leverage the retrieved contextual knowledge. We examine representative LVLMs and identify two major causes, namely, attention bias toward different tokens and knowledge conflicts between parametric and contextual knowledge. To this end, we design Adaptive Logits Fusion and Attention Reallocation (ALFAR), a training-free and plug-and-play approach that improves LVLM responses by maximizing the utility of the retrieved knowledge. Specifically, ALFAR tackles the challenges from two perspectives. First, it alleviates attention bias by adaptively shifting attention from visual tokens to relevant context tokens according to query-context relevance. Second, it decouples and weights parametric and contextual knowledge at output logits, mitigating conflicts between the two types of knowledge. As a plug-and-play method, ALFAR achieves superior performance across diverse datasets without requiring additional training or external tools. Extensive experiments over multiple LVLMs and benchmarks show that ALFAR consistently outperforms the state-of-the-art by large margins. Our code and data will be made publicly available.
Poster	Boosting Machine Unlearning via Ensembling Infinite Number of Functions on Task Arithmetic Simplex	https://neurips.cc//virtual/2025/poster/115936	Junhao Dong, Hao Zhu, Yifei Zhang, Xinghua Qu, Yew Soon Ong, Piotr Koniusz	As foundation Vision-Language Models (VLMs) unlock fine-tuning on smaller datasets while capturing large-scale data in pre-training, machine unlearning becomes critical in addressing privacy concerns and regulatory compliance. Task vector, representing differences between parameters of models fine-tuned with and without specific data, is a popular retraining-free unlearning strategy. However, we observe that task vectors exhibit substantial sensitivity to various fine-tuning configurations, resulting in unstable unlearning effectiveness negatively correlating with prediction-level variance. While aggregating multiple functions (e.g., VLM with classifier) whose parameters are represented by different task vectors naturally reduces function variance and improves unlearning, the computational cost of obtaining numerous task vectors and aggregating functions is computationally high. Thus, to robustly capture the space of task vectors induced by diverse fine-tuning strategies, we propose modeling it within the convex hull of $(Q-1)$-simplex whose vertices are $Q$ task vectors. Although a function ensemble can be formed by sampling numerous task vectors from such a simplex, this is computationally prohibitive. Thus, we derive a closed-form ensemble of an infinite number of functions whose parameters are uniformly sampled from the simplex, enabling efficient function-level task vector ensembling with enhanced unlearning performance. Extensive experiments and analyses across diverse datasets and scenarios demonstrate the efficacy of our method.
Poster	Boosting Multimodal Instance Understanding via Explicit Visual Prompt Instruction Tuning	https://neurips.cc//virtual/2025/poster/119587	Wujian Peng, Lingchen Meng, Yitong Chen, Yiweng Xie, Yang Liu, Tao Gui, Hang Xu, Xipeng Qiu, Zuxuan Wu, Yu-Gang Jiang	Large Multimodal Models (LMMs) have made significant breakthroughs with the advancement of instruction tuning. However, while existing models can understand images and videos at a holistic level, they still struggle with instance-level understanding that requires a more fine-grained comprehension and alignment. Instance-level understanding is crucial for LMMs, as it focuses on the specific elements that we are most interested in. Excitingly, existing works find that the state-of-the-art LMMs exhibit strong instance understanding capabilities when provided with explicit visual cues. Motivated by this, we proposed Inst-IT, a solution to enhance LMMs in Instance understanding via explicit visual prompt Instruction Tuning for instance guidance. Inst-IT consists of a benchmark to diagnose multimodal instance-level understanding, a large-scale instruction-tuning dataset, and a continuous instruction-tuning training paradigm to effectively enhance spatial-temporal instance understanding capabilities of existing LMMs. Experimental results show that, enhanced by Inst-IT, our models not only achieve outstanding performance on Inst-IT-Bench and other instance understanding benchmarks, but also demonstrate significant improvements across various generic image and video understanding benchmarks. This highlights that our method not only boosts instance-level understanding but also strengthens the overall capabilities of generic image and video comprehension.
Poster	Boosting Resilience of Large Language Models through Causality-Driven Robust Optimization	https://neurips.cc//virtual/2025/poster/116769	Xiaoling Zhou, Mingjie Zhang, Zhemg Lee, YUNCHENG HUA, chengli xing, Wei Ye, Flora Salim, Shikun Zhang	Large language models (LLMs) have achieved remarkable achievements across diverse applications; however, they remain plagued by spurious correlations and the generation of hallucinated content. Despite extensive efforts to enhance the resilience of LLMs, existing approaches either rely on indiscriminate fine-tuning of all parameters, resulting in parameter inefficiency and lack of specificity, or depend on post-processing techniques that offer limited adaptability and flexibility. This study introduces a novel Causality-driven Robust Optimization (CdRO) approach that selectively updates model components sensitive to causal reasoning, enhancing model causality while preserving valuable pretrained knowledge to mitigate overfitting. Our method begins by identifying the parameter components within LLMs that capture causal relationships, achieved through comparing the training dynamics of parameter matrices associated with the original samples, as well as augmented counterfactual and paraphrased variants. These comparisons are then fed into a lightweight logistic regression model, optimized in real time to dynamically identify and adapt the causal components within LLMs. The identified parameters are subsequently optimized using an enhanced policy optimization algorithm, where the reward function is designed to jointly promote both model generalization and robustness. Extensive experiments across various tasks using twelve different LLMs demonstrate the superior performance of our framework, underscoring its significant effectiveness in reducing the model’s dependence on spurious associations and mitigating hallucinations.
Poster	Boosting Skeleton-based Zero-Shot Action Recognition with Training-Free Test-Time Adaptation	https://neurips.cc//virtual/2025/poster/115304	Jingmin Zhu, Anqi Zhu, Hossein Rahmani, Jun Liu, Mohammed Bennamoun, Qiuhong Ke	We introduce \textit{Skeleton-Cache}, the first training-free test-time adaptation framework for skeleton-based zero-shot action recognition (SZAR), aimed at improving model generalization to unseen actions during inference. Skeleton-Cache reformulates inference as a lightweight retrieval process over a non-parametric cache that stores structured skeleton representations, combining both global and fine-grained local descriptors. To guide the fusion of descriptor-wise predictions, we leverage the semantic reasoning capabilities of large language models (LLMs) to assign class-specific importance weights. By integrating these structured descriptors with LLM-guided semantic priors, Skeleton-Cache dynamically adapts to unseen actions without any additional training or access to training data. Extensive experiments on NTU RGB+D 60/120 and PKU-MMD II demonstrate that Skeleton-Cache consistently boosts the performance of various SZAR backbones under both zero-shot and generalized zero-shot settings. Code will be released.
Poster	Boosting the Uniqueness of Neural Networks Fingerprints with Informative Triggers	https://neurips.cc//virtual/2025/poster/118065	Zhuomeng Zhang, Fangqi Li, Hanyi Wang, Shi-Lin Wang	One prerequisite for secure and reliable artificial intelligence services is tracing the copyright of backend deep neural networks. In the black-box scenario, the copyright of deep neural networks can be traced by their fingerprints, i.e., their outputs on a series of fingerprinting triggers. The performance of deep neural network fingerprints is usually evaluated in robustness, leaving the accuracy of copyright tracing among a large number of models with a lismited number of triggers intractable. This fact challenges the application of deep neural network fingerprints as the cost of queries is becoming a bottleneck. This paper studies the performance of deep neural network fingerprints from an information theoretical perspective. With this new perspective, we demonstrate that copyright tracing can be more accurate and efficient by using triggers with the largest marginal mutual information. Extensive experiments demonstrate that our method can be seamlessly incorporated into any existing fingerprinting scheme to facilitate the copyright tracing of deep neural networks.
Poster	Bootstrap Off-policy with World Model	https://neurips.cc//virtual/2025/poster/115054	Guojian Zhan, Likun Wang, Xiangteng Zhang, Jiaxin Gao, Masayoshi TOMIZUKA, Shengbo Li	Online planning has proven effective in reinforcement learning (RL) for improving sample efficiency and final performance. However, using planning for environment interaction inevitably introduces a divergence between the collected data and the policy's actual behaviors, degrading both model learning and policy improvement. To address this, we propose BOOM (Bootstrap Off-policy with WOrld Model), a framework that tightly integrates planning and off-policy learning through a bootstrap loop: the policy initializes the planner, and the planner refines actions to bootstrap the policy through behavior alignment. This loop is supported by a jointly learned world model, which enables the planner to simulate future trajectories and provides value targets to facilitate policy improvement. The core of BOOM is a likelihood-free alignment loss that bootstraps the policy using the planner’s non-parametric action distribution, combined with a soft value-weighted mechanism that prioritizes high-return behaviors and mitigates variability in the planner’s action quality within the replay buffer. Experiments on the high-dimensional DeepMind Control Suite and Humanoid-Bench show that BOOM achieves state-of-the-art results in both training stability and final performance.
Poster	Bootstrapping Hierarchical Autoregressive Formal Reasoner with Chain-of-Proxy-Autoformalization	https://neurips.cc//virtual/2025/poster/120138	Qi Liu, Xinhao Zheng, Renqiu Xia, Qinxiang Cao, Junchi Yan	Deductive formal problem-solving (D-FPS) enables process-verified, human-aligned problem-solving by implementing deductive solving processes within formal theorem proving (FTP) environments. However, current methods fail to address the misalignment between informal and formal reasoning granularity, and suffer from inefficiency due to backtracking and error propagation. Moreover, the extreme scarcity of formal problem-solution pairs further hinders progress.For the first gap, we propose HAR (_Hierarchical Autoregressive Formal Reasoner_), a novel reasoning pipeline. HAR decouples informal-aligned drafting and detailed proving, and formulates solution construction as autoregressive generation with per-step feedback. Second, we propose CoPA (_Chain-of-Proxy-Autoformalization_), a data generation pipeline that cascades statement autoformalization, proof drafting, and proof search as a proxy autoformalization path.Experiments demonstrate significant improvements: HAR achieves superior performance on FormalMath500 ($15.50\\% \mapsto 43.39\\%$) and MiniF2F-Solving ($21.87\\% \mapsto 55.68\\%$) with lower budget. CoPA shows consistent scalability through expert iteration. Explorations reveal promising directions in formal solution pruning and informal dataset denoising.
Poster	Bootstrap Your Uncertainty: Adaptive Robust Classification Driven by Optimal-Transport	https://neurips.cc//virtual/2025/poster/118390	Jiawei Huang, Minming Li, Hu Ding	Deep learning models often struggle with distribution shifts between training and deployment environments. Distributionally Robust Optimization (DRO) offers a promising framework by optimizing worst-case performance over a set of candidate distributions, which is called as the \emph{uncertainty set}. However, the efficacy of DRO heavily depends on the design of uncertainty set, and existing methods often perform suboptimally due to inappropriate and inflexible uncertainty sets. In this work, we first propose a novel perspective that casts entropy-regularized Wasserstein DRO as a dynamic process of distributional exploration and semantic alignment, both driven by optimal transport (OT). This unified viewpoint yields two key new techniques: \emph{semantic calibration}, which bootstraps semantically meaningful transport costs via inverse OT, and \emph{adaptive refinement}, which adjusts uncertainty set using OT-driven feedback. Together, these components form an exploration-and-feedback system, where the transport costs and uncertainty set evolve jointly during training, enabling the model to better adapt to potential distribution shifts. Moreover, we provide an in-depth analysis on this adaptive process and prove the theoretical convergence guarantee. Finally, we present our experimental results across diverse distribution shift scenarios, which demonstrate that our approach significantly outperforms existing methods, achieving state-of-the-art robustness.
Poster	Born a Transformer -- Always a Transformer?	https://neurips.cc//virtual/2025/poster/118861	Yana Veitsman, Mayank Jobanputra, Yash Sarrof, Aleksandra Bakalova, Vera Demberg, Ellie Pavlick, Michael Hahn	Transformers have theoretical limitations in modeling certain sequence-to-sequence tasks, yet it remains largely unclear if these limitations play a role in large-scale pretrained LLMs, or whether LLMs might effectively overcome these constraints in practice due to the scale of both the models themselves and their pretraining data. We explore how these architectural constraints manifest after pretraining, by studying a family of retrieval and copying tasks inspired by Liu et al. [2024]. We use the recently proposed C-RASP framework for studying length generalization [Huang et al., 2025b] to provide guarantees for each of our settings. Empirically, we observe an induction-versus-anti-induction asymmetry, where pretrained models are better at retrieving tokens to the right (induction) rather than the left (anti-induction) of a query token. This asymmetry disappears upon targeted fine-tuning if length-generalization is guaranteed by theory. Mechanistic analysis reveals that this asymmetry is connected to the differences in the strength of induction versus anti-induction circuits within pretrained Transformers. We validate our findings through practical experiments on real-world tasks demonstrating reliability risks. Our results highlight that pretraining selectively enhances certain Transformer capabilities, but does not overcome fundamental length-generalization limits.
Poster	Boundary-to-Region Supervision for Offline Safe Reinforcement Learning	https://neurips.cc//virtual/2025/poster/115428	Huikang Su, Dengyun Peng, Zifeng Zhuang, Yuhan Liu, Qiguang Chen, Donglin Wang, Qinghe Lli	Offline safe reinforcement learning aims to learn policies that satisfy predefined safety constraints from static datasets. Existing sequence-model-based methods condition action generation on symmetric input tokens for return-to-go and cost-to-go, neglecting their intrinsic asymmetry: RTG serves as a flexible performance target, while CTG should represent a rigid safety boundary. This symmetric conditioning leads to unreliable constraint satisfaction, especially when encountering out-of-distribution cost trajectories. To address this, we propose Boundary-to-Region (B2R), a framework that enables asymmetric conditioning through cost signal realignment . B2R redefines CTG as a boundary constraint under a fixed safety budget, unifying the cost distribution of all feasible trajectories while preserving reward structures. Combined with rotary positional embeddings , it enhances exploration within the safe region. Experimental results show that B2R satisfies safety constraints in 35 out of 38 safety-critical tasks while achieving superior reward performance over baseline methods. This work highlights the limitations of symmetric token conditioning and establishes a new theoretical and practical approach for applying sequence models to safe RL.
Poster	Boundary-Value PDEs Meet Higher-Order Differential Topology-aware GNNs	https://neurips.cc//virtual/2025/poster/118187	Yunfeng Liao, Yangxin Wu, Xiucheng Li	Recent advances in graph neural network (GNN)-based neural operators have demonstrated significant progress in solving partial differential equations (PDEs) by effectively representing computational meshes. However, most existing approaches overlook the intrinsic physical and topological meaning of higher-order elements in the mesh, which are closely tied to differential forms. In this paper, we propose a higher-order GNN framework that incorporates higher-order interactions based on discrete and finite element exterior calculus. The time-independent boundary value problems (BVPs) in electromagnetism are instantiated to illustrate the proposed framework. It can be easily generalized to other PDEs that admit differential form formulations. Moreover, the novel physics-informed loss terms, integrated form estimators, and theoretical support are derived correspondingly. Experiments show that our proposed method outperforms the existing neural operators by large margins on BVPs in electromagnetism. Our code is available at https://anonymous.4open.science/r/NeurIPS2025-18F5.
Poster	Bounds on the computational complexity of neurons due to dendritic morphology	https://neurips.cc//virtual/2025/poster/119616	Anamika Agrawal, Michael Buice	The simple linear threshold units used in many artificial neural networks have a limited computational capacity. Famously, a single unit cannot handle non-linearly separable problems like XOR. By contrast, real neurons exhibit complex morphologies as well as active dendritic integration, suggesting that their computational capacities outstrip those of simple linear units. Considering specific families of Boolean functions, we examine the computational limits of single units that incorporate more complex dendritic structures. For random Boolean functions, we show that there is a phase transition in learnability as a function of the input dimension, with most random functions below a certain critical dimension being learnable and those above not. This critical dimension is best predicted by the overall size of the dendritic arbor. This demonstrates real neurons have a far higher computational complexity than is usually considered in neural models, whether in machine learning or computational neuroscience. Furthermore, using architectures that are, respectively, more "apical" or "basal", we show that there are non-trivially disjoint sets of learnable functions by each type of neuron. Importantly, these two types of architectures differ in the robustness and generality of the computations they can perform. The basal-like architecture shows a higher probability of function realization, while the apical-like architecture shows an advantage with fast retraining for different functions. Given the cell-type specificity of morphological characteristics, these results suggest both that different components of the dendritic arbor as well as distinct cell types may have distinct computational roles. Our analysis offers new directions for neuron-level inductive biases in NeuroAI models using scalable models for neuronal cell-type specific computation.
Poster	BountyBench: Dollar Impact of AI Agent Attackers and Defenders on Real-World Cybersecurity Systems	https://neurips.cc//virtual/2025/poster/121456	Andy Zhang, Joey Ji, Celeste Menders, Riya Dulepet, Thomas Qin, Ron Wang, Junrong Wu, Kyleen Liao, Jiliang Li, Jinghan Hu, Sara Hong, Nardos Demilew, Shivatmica Murgai, Jason Tran, Nishka Kacheria, Ethan Ho, Denis Liu, Lauren McLane, Olivia Bruvik, Dai-Rong Han, Seungwoo Kim, Akhil Vyas, Cuiyuanxiu Chen, Ryan Li, Weiran Xu, Jonathan Ye, Prerit Choudhary, Siddharth M. Bhatia, Vikram Sivashankar, Yuxuan Bao, Dawn Song, Dan Boneh, Daniel Ho, Percy Liang	AI agents have the potential to significantly alter the cybersecurity landscape. To help us understand this change, we introduce the first framework to capture offensive and defensive cyber-capabilities in evolving real-world systems. Instantiating this framework with BountyBench, we set up 25 systems with complex, real-world codebases. To capture the vulnerability lifecycle, we define three task types: Detect (detecting a new vulnerability), Exploit (exploiting a given vulnerability), and Patch (patching a given vulnerability). For Detect, we construct a new success indicator, which is general across vulnerability types and provides localized evaluation. We manually set up the environment for each system, including installing packages, setting up server(s), and hydrating database(s). We add 40 bug bounties, which are vulnerabilities with monetary awards from \\$10 to \\$30,485, and cover 9 of the OWASP Top 10 Risks. To modulate task difficulty, we devise a new strategy based on information to guide detection, interpolating from identifying a zero day to exploiting a given vulnerability. We evaluate 5 agents: Claude Code, OpenAI Codex, and custom agents with GPT-4.1, Gemini 2.5 Pro Preview, and Claude 3.7 Sonnet Thinking. The top-performing agents are Claude Code (2.5% on Detect, corresponding to \\$450), Custom Agent with Claude 3.7 Sonnet Thinking (55% on Exploit), and OpenAI Codex (80% on Patch, corresponding to \\$13,710). The custom agents achieve higher Exploit scores of 35-55% compared to Patch scores of 30-35%; in contrast, OpenAI Codex and Claude Code achieve higher Patch scores of 80% and 57.5%, compared to Exploit scores of 25% and 32.5% respectively.
Poster	BRACE: A Benchmark for Robust Audio Caption Quality Evaluation	https://neurips.cc//virtual/2025/poster/121694	Tianyu Guo, Hongyu Chen, Hao Liang, Meiyi Qiang, Bohan Zeng, Linzhuang Sun, Bin CUI, Wentao Zhang	Automatic audio captioning is essential for audio understanding, enabling applications such as accessibility and content indexing. However, evaluating the quality of audio captions remains a major challenge, especially in reference-free settings where high-quality ground-truth captions are unavailable. While CLAPScore is currently the most widely used reference-free Audio Caption Evaluation Metric(ACEM), its robustness under diverse conditions has not been systematically validated. To address this gap, we introduce BRACE, a new benchmark designed to evaluate audio caption alignment quality in a reference-free setting. BRACE is primarily designed for assessing ACEMs, and can also be extended to measure the modality alignment abilities of Large Audio Language Model(LALM). BRACE consists of two sub-benchmarks: BRACE-Main for fine-grained caption comparison and BRACE-Hallucination for detecting subtle hallucinated content. We construct these datasets through high-quality filtering, LLM-based corruption, and human annotation. Given the widespread adoption of CLAPScore as a reference-free ACEM and the increasing application of LALMs in audio-language tasks, we evaluate both approaches using the BRACE benchmark, testing CLAPScore across various CLAP model variants and assessing multiple LALMs. Notably, even the best-performing CLAP-based ACEM achieves only a 70.01 F1-score on the BRACE-Main benchmark, while the best LALM reaches just 63.19. By revealing the limitations of CLAP models and LALMs, our BRACE benchmark offers valuable insights into the direction of future research. Our evaluation code and benchmark dataset are released in https://github.com/HychTus/BRACE_Evaluation and https://huggingface.co/datasets/gtysssp/audio_benchmarks.
Poster	BrainEC-LLM: Brain Effective Connectivity Estimation by Multiscale Mixing LLM	https://neurips.cc//virtual/2025/poster/116063	Wen Xiong, Junzhong Ji, Jinduo Liu	Pre-trained Large language models (LLMs) have shown impressive advancements in functional magnetic resonance imaging (fMRI) analysis and causal discovery. Considering the unique nature of the causal discovery field, which focuses on extracting causal graphs from observed data, research on LLMs in this field is still at an early exploratory stage. As a subfield of causal discovery, effective connectivity (EC) has received even less attention, and LLM-based approaches in EC remain unexplored. Existing LLM-based approaches for causal discovery typically rely on iterative querying to assess the causal influence between variable pairs, without any model adaptation or fine-tuning, making them ill-suited for handling the cross-modal gap and complex causal structures. To this end, we propose BrainEC-LLM, the first method to fine-tune LLMs for estimating brain EC from fMRI data. Specifically, multiscale decomposition mixing module decomposes fMRI time series data into short-term and long-term multiscale trends, then mixing them in bottom-up (fine to coarse) and top-down (coarse to fine) manner to extract multiscale temporal variations. And cross attention is applied with pre-trained word embeddings to ensure consistency between the fMRI input and pre-trained natural language. The experimental results on simulated and real resting-state fMRI datasets demonstrate that BrainEC-LLM can achieve superior performance when compared to state-of-the-art baselines.
Poster	BrainFlow: A Holistic Pathway of Dynamic Neural System on Mainfold	https://neurips.cc//virtual/2025/poster/116813	Zhixuan Zhou, Tingting Dan, Guorong Wu	A fundamental challenge in cognitive neuroscience is understanding how cognition emerges from the interplay between structural connectivity (SC) and dynamic functional connectivity (FC) in the brain. Network neuroscience has emerged as a powerful framework to understand brain function through a holistic perspective on structure-function relationships. In this context, current machine learning approaches typically seek to establish direct mappings between structural connectivity (SC) and functional connectivity (FC) associated with specific cognitive states.However, these state-independent methods often yield inconsistent results due to overlapping brain networks across cognitive states. To address this limitation, we conceptualize to uncover the dendritic coupling mechanism between one static SC and multiple FCs by solving a flow problem that bridges the distribution of SC to a mixed distribution of FCs, conditioned on various cognitive states, along a Riemannian manifold of symmetric positive-definite (SPD) manifold.We further prove the equivalence between flow matching on the SPD manifold and on the computationally efficient Cholesky manifold. Since a spare of functional connections is shared across cognitive states, we introduce the notion of consensus control to promote the shared kinetic structures between multiple FC-to-SC pathways via synchronized coordination, yielding a biologically meaningful underpinning on SC-FC coupling mechanism.Together, we present BrainFlow, a reversible generative model that achieves state-of-the-art performance on not only synthetic data but also large-scale neuroimaging datasets from UK Biobank and Human Connectome Project.
Poster	Brain Harmony: A Multimodal Foundation Model Unifying Morphology and Function into 1D Tokens	https://neurips.cc//virtual/2025/poster/115597	Zijian Dong, Li Ruilin, Joanna Chong, Niousha Dehestani, Yinghui Teng, Yi Lin, Zhizhou Li, Yichi Zhang, Yapei Xie, Leon Ooi, B.T. Yeo, Juan Helen Zhou	We present Brain Harmony (BrainHarmonix), the first multimodal brain foundation model that unifies structural morphology and functional dynamics into compact 1D token representations. The model was pretrained on two of the largest neuroimaging datasets to date, encompassing 64,594 T1-weighted structural MRI 3D volumes (~14 million images) and 70,933 functional MRI (fMRI) time series. BrainHarmonix is grounded in two foundational neuroscience principles: structure complements function - structural and functional modalities offer distinct yet synergistic insights into brain organization; function follows structure - brain functional dynamics are shaped by cortical morphology. The modular pretraining process involves single-modality training with geometric pre-alignment followed by modality fusion through shared brain hub tokens. Notably, our dynamics encoder uniquely handles fMRI time series with heterogeneous repetition times (TRs), addressing a major limitation in existing models. BrainHarmonix is also the first to deeply compress high-dimensional neuroimaging signals into unified, continuous 1D tokens, forming a compact latent space of the human brain. BrainHarmonix achieves strong generalization across diverse downstream tasks, including neurodevelopmental and neurodegenerative disorder classification and cognition prediction - consistently outperforming previous approaches. Our models - pretrained on 8 H100 GPUs - will be made publicly available, aiming to catalyze a new era of AI-driven neuroscience powered by large-scale multimodal neuroimaging.
Poster	Brain-Informed Fine-Tuning for Improved Multilingual Understanding in Language Models	https://neurips.cc//virtual/2025/poster/118728	Anuja Negi, SUBBAREDDY OOTA, Anwar Nunez-Elizalde, Manish Gupta, Fatma Deniz	Recent studies have demonstrated that fine-tuning language models with brain data can improve their semantic understanding, although these findings have so far been limited to English. Interestingly, similar to the shared multilingual embedding space of pretrained multilingual language models, human studies provide strong evidence for a shared semantic system in bilingual individuals. Here, we investigate whether fine-tuning language models with bilingual brain data changes model representations in a way that improves them across multiple languages. To test this, we fine-tune monolingual and multilingual language models using brain activity recorded while bilingual participants read stories in English and Chinese. We then evaluate how well these representations generalize to the bilingual participants’ first language, their second language, and several other languages that the participant is not fluent in. We assess the fine-tuned language models on brain encoding performance and downstream NLP tasks. Our results show that bilingual brain-informed fine-tuned language models outperform their vanilla (pretrained) counterparts in both brain encoding performance and most downstream NLP tasks across multiple languages. These findings suggest that brain-informed fine-tuning improves multilingual understanding in language models, offering a bridge between cognitive neuroscience and NLP research.
Poster	Brain-Like Processing Pathways Form in Models With Heterogeneous Experts	https://neurips.cc//virtual/2025/poster/118100	Jack Cook, Danyal Akarca, Rui Costa, Jascha Achterberg	The brain is made up of a vast set of heterogeneous regions that dynamically organize into pathways as a function of task demands. Examples of such pathways can be seen in the interactions between cortical and subcortical networks during learning. This raises the question of how exactly brain regions organize into these dynamic groups. In this work, we use an extension of the Heterogeneous Mixture-of-Experts architecture, to show that heterogeneous regions do not form processing pathways by themselves, implying that the brain likely implements specific constraints which result in reliable formation of pathways. We identify three biologically relevant inductive biases that encourage pathway formation: a routing cost imposed on the use of more complex regions, a scaling factor that reduces this cost when task performance is low, and randomized expert dropout. When comparing our resulting Mixture-of-Pathways model with the brain, we observe that the artificial pathways match how the brain uses cortical and subcortical systems to learn and solve tasks of varying difficulty. In summary, we introduce a novel framework for investigating how the brain forms task-specific pathways through inductive biases which may make Mixture-of-Experts architectures in general more adaptive.
Poster	Brain-like variational inference	https://neurips.cc//virtual/2025/poster/119894	Hadi Vafaii, Dekel Galor, Jacob Yates	Inference in both brains and machines can be formalized by optimizing a shared objective: maximizing the evidence lower bound (ELBO) in machine learning, or minimizing variational free energy ($\mathcal{F}$) in neuroscience (ELBO = $-\mathcal{F}$). While this equivalence suggests a unifying framework, it leaves open how inference is implemented in neural systems. Here, we show that online natural gradient descent on $\mathcal{F}$, under Poisson assumptions, leads to a recurrent spiking neural network that performs variational inference via membrane potential dynamics. The resulting model—the iterative Poisson variational autoencoder (iP-VAE)—replaces the encoder network with local updates derived from natural gradient descent on $\mathcal{F}$. Theoretically, iP-VAE yields a number of desirable features such as emergent normalization via lateral competition, and hardware-efficient integer spike count representations. Empirically, iP-VAE outperforms both standard VAEs and Gaussian-based predictive coding models in sparsity, reconstruction, and biological plausibility. iP-VAE also exhibits strong generalization to out-of-distribution inputs, exceeding hybrid iterative-amortized VAEs. These results demonstrate how deriving inference algorithms from first principles can yield concrete architectures that are simultaneously biologically plausible and empirically effective.
Poster	BrainMoE: Cognition Joint Embedding via Mixture-of-Expert Towards Robust Brain Foundation Model	https://neurips.cc//virtual/2025/poster/120350	Ziquan Wei, Tingting Dan, Tianlong Chen, Guorong Wu	Given the large scale of public functional Magnetic Resonance Imaging (fMRI), e.g., UK Biobank (UKB) and Human Connectome Projects (HCP), brain foundation models are emerging. Although the amount of samples under rich environmental variables is unprecedented, existing brain foundation models learn from fMRI derived from a narrow range of cognitive states stimulated by similar environments, causing the limited robustness demonstrated in various applications and datasets acquired with different pipelines and limited sample size. By capitalizing on the variety of cognitive status as subjects performing explicit tasks, we present the mixture of brain experts, namely BrainMoE, pre-training on tasking fMRI with rich behavioral tasks in addition to resting fMRI for a robust brain foundation model. Brain experts are designed to produce embeddings for different behavioral tasks related to cognition. Afterward, these cognition embeddings are mixed by a cognition adapter via cross-attention so that BrainMoE can handle orthogonal embeddings and be robust on those boutique downstream datasets. We have pre-trained two existing self-regressive architectures and one new supervised architecture as brain experts on 68,251 fMRI scans among UKB and HCP, containing 12 different cognitive states. Then, BrainMoE is evaluated on a variety of applications, including sex prediction, human behavior recognition, and disease early diagnosis of Autism, Parkinson's disease, Alzheimer's disease, and Schizophrenia, where promising results in seven datasets from three different pipelines indicate great potential to facilitate current neuroimaging applications in clinical routines.
Poster	Brain network science modelling of sparse neural networks enables Transformers and LLMs to perform as fully connected	https://neurips.cc//virtual/2025/poster/118300	Yingtao Zhang, Diego Cerretti, Jialin Zhao, Wenjing Wu, Ziheng Liao, Umberto Michieli, Carlo Cannistraci	This study aims to enlarge our current knowledge on the application of brain-inspired network science principles for training artificial neural networks (ANNs) with sparse connectivity. Dynamic sparse training (DST) emulates the synaptic turnover of real brain networks, reducing the computational demands of training and inference in ANNs. However, existing DST methods face difficulties in maintaining peak performance at high connectivity sparsity levels. The Cannistraci-Hebb training (CHT) is a brain-inspired method that is used in DST for growing synaptic connectivity in sparse neural networks. CHT leverages a gradient-free, topology-driven link regrowth mechanism, which has been shown to achieve ultra-sparse (1\% connectivity or lower) advantage across various tasks compared to fully connected networks. Yet, CHT suffers two main drawbacks: (i) its time complexity is $\mathcal{O}(N\cdot d^3)$- N node network size, d node degree - hence it can be efficiently applied only to ultra-sparse networks. (ii) it rigidly selects top link prediction scores, which is inappropriate for the early training epochs, when the network topology presents many unreliable connections. Here, we design the first brain-inspired network model - termed bipartite receptive field (BRF) - to initialize the connectivity of sparse artificial neural networks. Then, we propose a matrix multiplication GPU-friendly approximation of the CH link predictor, which reduces the computational complexity to $\mathcal{O}(N^3)$, enabling a fast implementation of link prediction in large-scale models. Moreover, we introduce the Cannistraci-Hebb training soft rule (CHTs), which adopts a flexible strategy for sampling connections in both link removal and regrowth, balancing the exploration and exploitation of network topology. To further enhance performance, we integrate CHTs with a brain-inspired network topology initialization method known as the bipartite receptive field (BRF). Additionally, we propose a sigmoid-based gradual density decay strategy, leading to an advanced framework referred to as CHTss. Empirical results show that BRF offers performance advantages over previous network science models. Using 1\% of connections, CHTs outperforms fully connected networks in MLP architectures on visual classification tasks, compressing some networks to less than 30\% of the nodes. Using 5\% of the connections, CHTss outperforms fully connected networks in two Transformer-based machine translation tasks. Finally, using 30\% of the connections, CHT and CHTss achieve superior performance compared to other dynamic sparse training methods in language modeling across different sparsity levels, and it surpasses the fully connected counterpart in zero-shot evaluations.
Poster	BrainODE: Neural Shape Dynamics for Age- and Disease-aware Brain Trajectories	https://neurips.cc//virtual/2025/poster/119662	Wonjung Park, Suhyun Ahn, Maria Hernandez, Susana Maniega, Jinah Park	We propose BrainODE, a neural ordinary differential equations (ODEs)-based framework for modeling continuous longitudinal deformations of brain shapes. BrainODE learns a deformation space over anatomically meaningful brain regions to enable early prediction of neurodegenerative disease progression. Addressing key challenges in longitudinal neuroimaging— including limited data, irregular sampling intervals, and inter-subject variability—we design a conditional neural ODE architecture that modulates shape dynamics with subject-specific age and cognitive status. To enable autoregressive forecasting from a single observation, we introduce a pseudo-cognitive status embedding that jointly encodes age and estimated cognitive status. We demonstrate that BrainODE outperforms time-aware baselines in predicting future brain shapes, demonstrating strong generalization across longitudinal datasets with both regular and irregular time intervals.
Poster	BrainOmni: A Brain Foundation Model for Unified EEG and MEG Signals	https://neurips.cc//virtual/2025/poster/117066	Qinfan Xiao, Ziyun Cui, Chi Zhang, SiQi Chen, Wen Wu, Andrew Thwaites, Alexandra Woolgar, Bowen Zhou, Chao Zhang	Electroencephalography (EEG) and magnetoencephalography (MEG) measure neural activity non-invasively by capturing electromagnetic fields generated by dendritic currents. Although rooted in the same biophysics, EEG and MEG exhibit distinct signal patterns, further complicated by variations in sensor configurations across modalities and recording devices.Existing approaches typically rely on separate, modality- and dataset-specific models, which limits the performance and cross-domain scalability.This paper proposes BrainOmni, the first brain foundation model that generalises across heterogeneous EEG and MEG recordings. To unify diverse data sources, we introduce BrainTokenizer, the first tokeniser that quantises spatiotemporal brain activity into discrete representations. Central to BrainTokenizer is a novel Sensor Encoder that encodes sensor properties such as spatial layout, orientation, and type, enabling compatibility across devices and modalities.Building upon the discrete representations, BrainOmni learns unified semantic embeddings of brain signals by self-supervised pretraining. To the best of our knowledge, it is the first foundation model to support both EEG and MEG signals, as well as the first to incorporate large-scale MEG pretraining.A total of 1,997 hours of EEG and 656 hours of MEG data are curated and standardised from publicly available sources for pretraining.Experiments show that BrainOmni outperforms both existing foundation models and state-of-the-art task-specific models on a range of downstream tasks. It also demonstrates strong generalisation to unseen EEG and MEG devices. Further analysis reveals that joint EEG-MEG (EMEG) training yields consistent improvements across both modalities. Code and model checkpoints will be released upon acceptance.
Poster	Brain-Predictive Reasoning Embedding through Residual Disentanglement	https://neurips.cc//virtual/2025/poster/115596	Linyang He, Tianjun Zhong, Richard Antonello, Gavin Mischler, Micah Goldblum, Nima Mesgarani	Conventional brain encoding analysis using language models that feeds whole hidden states can be biased toward shallow lexical cues. Here we present a residual-layer disentangling method that extracts four nearly orthogonal vectors from a language model, respectively containing information corresponding to lexicon, syntax, meaning, and reasoning. We first probe the model to locate the layers where each linguistic feature is maximal, then strip lower-level feature incrementally. Applying bootstrap-ridge encoding to natural-speech ECoG yields three insights: 1) Our residual pipeline isolates a reasoning embedding with unique predictive value, possible only because the latest large language models exhibit emergent reasoning behavior. 2) Apparent high-level predictive performance in conventional analyses is largely attributable to recycled shallow information, rather than genuine deep processing. 3) The reasoning embedding reveals distinct spatiotemporal brain activation patterns, including recruitment of frontal and visual regions beyond classical language areas, suggesting a potential neural substrate for high-level reasoning. Together, our approach removes shallow bias, aligns distinct transformer strata with brain hierarchies, and provides the first brain-relevant representation of reasoning.
Poster	Brain-tuning Improves Generalizability and Efficiency of Brain Alignment in Speech Models	https://neurips.cc//virtual/2025/poster/119924	Omer Moussa, Mariya Toneva	Pretrained language models are remarkably effective in aligning with human brain responses (e.g., fMRI) elicited by natural language stimuli, positioning them as promising model organisms for studying language processing in the brain. However, existing approaches for both estimating and improving this brain alignment are participant-dependent and highly affected by the amount of data available per participant, hindering both generalization to new participants and population-level analyses. In this work, we address these limitations by introducing a scalable, generalizable brain-tuning method, in which we fine-tune pretrained speech language models to jointly predict fMRI responses from multiple participants who listen to the same natural stories. We demonstrate that the resulting brain-tuned models exhibit strong individual brain alignment while generalizing across participants. Specifically, our method leads to 1) a 5-fold decrease in the amount of fMRI data needed to predict brain data from new participants and 2) up to a 50\% increase in the overall brain alignment. Furthermore, multi-brain-tuning additionally improves downstream performance on semantic tasks, suggesting that training using brain data from multiple participants leads to more generalizable semantic representations. Taken together, these findings demonstrate a bidirectional benefit between neuroscience and AI, helping bridge the gap between the two fields.
Poster	BraVE: Offline Reinforcement Learning for Discrete Combinatorial Action Spaces	https://neurips.cc//virtual/2025/poster/118272	Matthew Landers, Taylor Killian, Hugo Barnes, Tom Hartvigsen, Afsaneh Doryab	Offline reinforcement learning in high-dimensional, discrete action spaces is challenging due to the exponential scaling of the joint action space with the number of sub-actions and the complexity of modeling sub-action dependencies. Existing methods either exhaustively evaluate the action space, making them computationally infeasible, or factorize Q-values, failing to represent joint sub-action effects. We propose \textbf{Bra}nch \textbf{V}alue \textbf{E}stimation (BraVE), a value-based method that uses tree-structured action traversal to evaluate a linear number of joint actions while preserving dependency structure. BraVE outperforms prior offline RL methods by up to $20\times$ in environments with over four million actions.
Poster	BREAD: Enhancing SLM Reasoning by Bridging Supervised and Reinforcement Learning	https://neurips.cc//virtual/2025/poster/118358	Xuechen Zhang, Zijian Huang, Yingcong Li, Chenshun Ni, Jiasi Chen, Samet Oymak	Small language models (SLMs) struggle to learn complex reasoning behaviors, especially when high-quality traces are scarce or difficult to learn from. A typical approach for training such models combines a supervised fine-tuning (SFT) stage, often to distill reasoning capabilities from a larger model, followed by a reinforcement learning (RL) stage such as Group Relative Policy Optimization (GRPO). In this paper, we investigate the fundamental limitations of this SFT + RL paradigm and propose methods to overcome them. Using a toy student-expert model over Markov chains, we demonstrate that the SFT + RL strategy can fail completely when (1) the expert's traces are too difficult for the small model to express, or (2) the small model's initialization achieves exponentially sparse rewards as task complexity grows. To address these, we introduce BREAD, a GRPO variant that bridges SFT and RL via partial expert guidance and branch rollouts. When self-generated traces fail, BREAD adaptively inserts short expert prefixes/hints, allowing the small model to complete the rest of the reasoning path, and ensuring that each update includes at least one successful trace. This mechanism both densifies the reward signal and induces a natural learning curriculum. BREAD requires fewer than 40\% of ground-truth traces, consistently outperforming standard GRPO while speeding up the training by about 3$\times$. Importantly, we find that BREAD helps the model solve problems that are otherwise unsolvable by the SFT + RL strategy, highlighting how branch rollouts and expert guidance can aid SLM reasoning.
Poster	Breaking Latent Prior Bias in Detectors for Generalizable AIGC Image Detection	https://neurips.cc//virtual/2025/poster/115689	Yue Zhou, Xinan He, Kaiqing Lin, Bing Fan, Feng Ding, Bin Li	Current AIGC detectors often achieve near-perfect accuracy on images produced by the same generator used for training but struggle to generalize to outputs from unseen generators. We trace this failure in part to latent prior bias: detectors learn shortcuts tied to patterns stemming from the initial noise vector rather than learning robust generative artifacts. To address this, we propose \textbf{On-Manifold Adversarial Training (OMAT)}: by optimizing the initial latent noise of diffusion models under fixed conditioning, we generate \emph{on-manifold} adversarial examples that remain on the generator’s output manifold—unlike pixel-space attacks, which introduce off-manifold perturbations that the generator itself cannot reproduce and that can obscure the true discriminative artifacts. To test against state-of-the-art generative models, we introduce GenImage++, a test-only benchmark of outputs from advanced generators (Flux.1, SD3) with extended prompts and diverse styles. We apply our adversarial-training paradigm to ResNet50 and CLIP baselines and evaluate across existing AIGC forensic benchmarks and recent challenge datasets. Extensive experiments show that adversarially trained detectors significantly improve cross-generator performance without any network redesign. Our findings on latent-prior bias offer valuable insights for future dataset construction and detector evaluation, guiding the development of more robust and generalizable AIGC forensic methodologies.
Poster	Breaking the Batch Barrier (B3) of Contrastive Learning via Smart Batch Mining	https://neurips.cc//virtual/2025/poster/118493	Raghuveer Thirukovalluru, Rui Meng, Ye Liu, Karthikeyan K, Mingyi Su, Ping Nie, Semih Yavuz, Yingbo Zhou, Wenhu Chen, Bhuwan Dhingra	Contrastive learning (CL) is a prevalent technique for training embedding models, which pulls semantically similar examples (positives) closer in the representation space while pushing dissimilar ones (negatives) further apart. A key source of negatives are 'in-batch' examples, i.e., positives from other examples in the batch. Effectiveness of such models is hence strongly influenced by the size and quality of training batches. In this work, we propose 'Breaking the Batch Barrier' (B3), a novel batch construction strategy designed to curate high-quality batches for CL. Our approach begins by using a pretrained teacher embedding model to rank all examples in the dataset, from which a sparse similarity graph is constructed. A community detection algorithm is then applied to this graph to identify clusters of examples that serve as strong negatives for one another. The clusters are then used to construct batches that are rich in in-batch negatives. Empirical results on the MMEB multimodal embedding benchmark (36 tasks) demonstrate that our method sets a new state of the art, outperforming previous best methods by +1.3 and +2.9 points at the 7B and 2B model scales, respectively. Notably, models trained with \bthm\ surpass existing state-of-the-art results even with a batch size as small as 64, which is 4–16× smaller than that required by other methods.
Poster	Breaking the Compression Ceiling: Data-Free Pipeline for Ultra-Efficient Delta Compression	https://neurips.cc//virtual/2025/poster/116137	Xiaohui Wang, Peng Ye, Chenyu Huang, Shenghe Zheng, Bo Zhang, LEI BAI, Wanli Ouyang, Tao Chen	With the rise of the fine-tuned–pretrained paradigm, storing numerous fine-tuned models for multi-tasking creates significant storage overhead.Delta compression alleviates this by storing only the pretrained model and the highly compressed delta weights (the differences between fine-tuned and pretrained model weights). However, existing methods fail to maintain both high compression and performance, and often rely on data.To address these challenges, we propose UltraDelta, the first data-free delta compression pipeline that achieves both ultra-high compression and strong performance. UltraDelta is designed to minimize redundancy, maximize information, and stabilize performance across inter-layer, intra-layer, and global dimensions, using three key components:(1) Variance-Based Mixed Sparsity Allocation assigns sparsity based on variance, giving lower sparsity to high-variance layers to preserve inter-layer information.(2) Distribution-Aware Compression applies uniform quantization and then groups parameters by value, followed by group-wise pruning, to better preserve intra-layer distribution.(3) Trace-Norm-Guided Rescaling uses the trace norm of delta weights to estimate a global rescaling factor, improving model stability under higher compression.Extensive experiments across (a) large language models (fine-tuned on LLaMA-2 7B and 13B) with up to 133$\times$, (b) general NLP models (RoBERTa-base, T5-base) with up to 800$\times$,(c) vision models (ViT-B/32, ViT-L/14) with up to 400$\times$, and(d) multi-modal models (BEiT-3) with 40$\times$ compression ratio, demonstrate that UltraDelta consistently outperforms existing methods, especially under ultra-high compression.
Poster	Breaking the Discretization Barrier of Continuous Physics Simulation Learning	https://neurips.cc//virtual/2025/poster/115234	Fan Xu, Hao Wu, Nan Wang, Lilan Peng, Kun Wang, Wei Gong, Xibin Zhao	The modeling of complicated time-evolving physical dynamics from partial observations is a long-standing challenge. Particularly, observations can be sparsely distributed in a seemingly random or unstructured manner, making it difficult to capture highly nonlinear features in a variety of scientific and engineering problems. However, existing data-driven approaches are often constrained by fixed spatial and temporal discretization. While some researchers attempt to achieve spatio-temporal continuity by designing novel strategies, they either overly rely on traditional numerical methods or fail to truly overcome the limitations imposed by discretization. To address these, we propose CoPS, a purely data-driven methods, to effectively model continuous physics simulation from partial observations. Specifically, we employ multiplicative filter network to fuse and encode spatial information with the corresponding observations. Then we customize geometric grids and use message-passing mechanism to map features from original spatial domain to the customized grids. Subsequently, CoPS models continuous-time dynamics by designing multi-scale graph ODEs, while introducing a Marcov-based neural auto-correction module to assist and constrain the continuous extrapolations. Comprehensive experiments demonstrate that CoPS advances the state-of-the-art methods in space-time continuous modeling across various scenarios. Our codes are available at~\url{https://anonymous.4open.science/r/CoPS-F625}.
Poster	Breaking the Frozen Subspace: Importance Sampling for Low-Rank Optimization in LLM Pretraining	https://neurips.cc//virtual/2025/poster/117335	Haochen Zhang, Junze Yin, Guanchu Wang, Zirui Liu, Lin Yang, Tianyi Zhang, Anshumali Shrivastava, Vladimir Braverman	Low-rank optimization has emerged as a promising approach to enabling memory-efficient training of large language models (LLMs). Existing low-rank optimization methods typically project gradients onto a low-rank subspace, reducing the memory cost of storing optimizer states. A key challenge in these methods is selecting suitable subspaces to ensure an effective optimization trajectory. Most existing approaches select the dominant subspace to preserve gradient information, as this intuitively provides the best approximation. However, we find that in practice, the dominant subspace stops changing during pretraining, thereby constraining weight updates to similar subspaces. In this paper, we propose importance sampling for low-rank optimization in LLM pretraining with a provable convergence guarantee, which the dominant subspace approach does not have. Empirically, we demonstrate that our method significantly outperforms previous methods in LLM pretraining tasks.
Poster	Breaking the Gold Standard: Extracting Forgotten Data under Exact Unlearning in Large Language Models	https://neurips.cc//virtual/2025/poster/119345	Xiaoyu Wu, Yifei Pang, Terrance Liu, Steven Wu	Large language models are typically trained on datasets collected from the web, which may inadvertently contain harmful or sensitive personal information. To address growing privacy concerns, unlearning methods have been proposed to remove the influence of specific data from trained models. Of these, exact unlearning---which retrains the model from scratch without the target data---is widely regarded the gold standard, believed to be robust against privacy-related attacks. In this paper, we challenge this assumption by introducing a novel data extraction attack that compromises even exact unlearning. Our method leverages both the pre- and post-unlearning models: by guiding the post-unlearning model using signals from the pre-unlearning model, we uncover patterns that reflect the removed data distribution. Combining model guidance with a token filtering strategy, our attack significantly improves extraction success rates---doubling performance in some cases---across common benchmarks such as MUSE, TOFU, and WMDP. Furthermore, we demonstrate our attack's effectiveness on a simulated medical diagnosis dataset to highlight real-world privacy risks associated with exact unlearning. In light of our findings, which suggest that unlearning may, in a contradictory way, increase the risk of privacy leakage, we advocate for evaluation of unlearning methods to consider broader threat models that account not only for post-unlearning models but also for adversarial access to prior checkpoints.
Poster	Breaking the Gradient Barrier: Unveiling Large Language Models for Strategic Classification	https://neurips.cc//virtual/2025/poster/117729	Xinpeng Lv, Haotian Wang, Yunxin Mao, KE LIANG, Haoxuan Li, Wanrong Huang, Long Lan, Haoang Chi, Huan Chen, Jinxuan Yang, Cyuanlong, Wenjing Yang	Strategic classification (SC) explores how individuals or entities modify their features strategically to achieve favorable classification outcomes. However, existing SC methods, which are largely based on linear models or shallow neural networks, face significant limitations in terms of scalability and capacity when applied to real-world datasets with significantly increasing scale, especially in financial services and the internet sector. In this paper, we investigate how to leverage large language models to design a more scalable and efficient SC framework, especially in the case of growing individuals engaged with decision-making processes. Specifically, we introduce GLIM, a gradient-free SC method grounded in in-context learning. During the feed-forward process of self-attention, GLIM implicitly simulates the typical bi-level optimization process of SC, including both the feature manipulation and decision rule optimization. Without fine-tuning the LLMs, our proposed GLIM enjoys the advantage of cost-effective adaptation in dynamic strategic environments. Theoretically, we prove GLIM can support pre-trained LLMs to adapt to a broad range of strategic manipulations. We validate our approach through experiments with a bunch of pre-trained LLMs on real-world and synthetic datasets in financial and internet domains, demonstrating that our GLIM exhibits both robustness and efficiency, and offering an effective solution for large-scale SC tasks.
Poster	Breaking the Order Barrier: Off-Policy Evaluation for Confounded POMDPs	https://neurips.cc//virtual/2025/poster/115637	Qi Kuang, Jiayi Wang, Fan Zhou, Zhengling Qi	We consider off-policy evaluation (OPE) in Partially Observable Markov Decision Processes (POMDPs) with unobserved confounding. Recent advances have introduced bridge-function to circumvent unmeasured confounding and develop estimators for the policy value, yet the statistical error bounds of them related to the length of horizon $T$ and the size of the state-action space $\|\mathcal{O}\|\|\mathcal{A}\|$ remain largely unexplored. In this paper, we systematically investigate finite-sample error bounds of OPE estimators in finite-horizon tabular confounded POMDPs. Specifically, we show that under certain rank conditions, the estimation error for policy value can achieve a rate of $\mathcal{O}(T^{1.5}/\sqrt{n})$, excluding the cardinality of observation space $\|\mathcal{O}\|$ and action space $\|\mathcal{A}\|$. With an additional and mild condition on the concentrability coefficients in confounded POMDPs, the rate of estimation error can be improved to $\mathcal{O}(T/\sqrt{n})$. We also show that for fully history-dependent policy, the estimation error scales as $\mathcal{O}\big(T/\sqrt{n}(\|\mathcal{O}\| \|\mathcal{A}\|)^{\frac{T}{2}}\big)$, highlighting the exponential error dependence introduced by history-based proxies to infer hidden states. Furthermore, when the target policy is memoryless policy, the error bound improves to $\mathcal{O}\big(T/\sqrt{n}\sqrt{\|\mathcal{O}\| \|\mathcal{A}\|}\big)$, which matches the optimal rate known for tabular MDPs. To the best of our knowledge, this is the first work to provide a comprehensive finite-sample analysis of OPE in confounded POMDPs.
Poster	Breaking the Performance Ceiling in Complex Reinforcement Learning requires Inference Strategies	https://neurips.cc//virtual/2025/poster/117975	Felix Chalumeau, Daniel Rajaonarivonivelomanantsoa, Ruan John de Kock, Juan Formanek, Sasha Abramowitz, Omayma Mahjoub, Wiem Khlifi, Simon Du Toit, Louay Nessir, Refiloe Shabe, Arnol Fokam, Siddarth Singh, Ulrich Armel Mbou Sob, Arnu Pretorius	Reinforcement learning (RL) systems have countless applications, from energy-grid management to protein design. However, such real-world scenarios are often extremely difficult, combinatorial in nature, and require complex coordination between multiple agents. This level of complexity can cause even state-of-the-art RL systems, trained until convergence, to hit a performance ceiling which they are unable to break out of with zero-shot inference. Meanwhile, many digital or simulation-based applications allow for an inference phase that utilises a specific time and compute budget to explore multiple attempts before outputting a final solution. In this work, we show that such an inference phase employed at execution time, and the choice of a corresponding inference strategy, are key to breaking the performance ceiling observed in complex multi-agent RL problems. Our main result is striking: we can obtain up to a 126% and, on average, a 45% improvement over the previous state-of-the-art across 17 tasks, using only a couple seconds of extra wall-clock time during execution. We also demonstrate promising compute scaling properties, supported by over 60k experiments, making it the largest study on inference strategies for complex RL to date. We make all of our experimental data and code available.
Poster	Breakthrough Sensor-Limited Single View: Towards Implicit Temporal Dynamics for Time Series Domain Adaptation	https://neurips.cc//virtual/2025/poster/117174	Mingyang Liu, Xinyang Chen, Xiucheng Li, Weili Guan, Liqiang Nie	Unsupervised domain adaptation has emerged as a pivotal paradigm for mitigating distribution shifts in time series analysis. The fundamental challenge in time series domain adaptation arises from the entanglement of domain shifts and intricate temporal patterns. Crucially, the latent continuous-time dynamics, which are often inaccessible due to sensor constraints, are only partially observable through discrete time series from an explicit sensor-limited single view. This partial observability hinders the modeling of intricate temporal patterns, impeding domain invariant representation learning. To mitigate the limitation, we propose EDEN (multiple Explicit Domain Enhanced adaptation Network), expanding the raw dataset to multi-scale explicit domains, multi-subspace explicit domains and multi-segment explicit domains. EDEN enhances domain adaptation with three coordinated modules tailored to integrate multiple explicit domains: (1) Multi-Scale Curriculum Adaptation implements progressive domain alignment from coarse-scale to fine-scale. (2) Quality-Aware Feature Fusion evaluates feature quality in multi-subspace explicit domains and adaptively integrates temporal-frequency features. (3) Temporal Coherence Learning enforces segment-level consistency with multi-segment explicit domains. The representation enriched by multiple explicit domains bridges the gap between partially observed discrete samples and the underlying implicit temporal dynamics, enabling more accurate approximation of implicit temporal patterns for effective cross-domain adaptation. Our comprehensive evaluation across 6 time series benchmarks demonstrates EDEN's consistent superiority, achieving average accuracy improvements of 4.8% over state-of-the-art methods in cross-domain scenarios. Code is available at the anonymous link: <https://anonymous.4open.science/r/2025NeurIPS-EDEN>.
Poster	BridgePure: Limited Protection Leakage Can Break Black-Box Data Protection	https://neurips.cc//virtual/2025/poster/118703	Yihan Wang, Yiwei Lu, Xiao-Shan Gao, Gautam Kamath, Yaoliang Yu	Availability attacks, or unlearnable examples, are defensive techniques that allow data owners to modify their datasets in ways that prevent unauthorized machine learning models from learning effectively while maintaining the data's intended functionality. It has led to the release of popular black-box tools (e.g., APIs) for users to upload personal data and receive protected counterparts. In this work, we show that such black-box protections can be substantially compromised if a small set of unprotected in-distribution data is available. Specifically, we propose a novel threat model of protection leakage, where an adversary can (1) easily acquire (unprotected, protected) pairs by querying the black-box protections with a small unprotected dataset; and (2) train a diffusion bridge model to build a mapping between unprotected and protected data. This mapping, termed BridgePure, can effectively remove the protection from any previously unseen data within the same distribution. BridgePure demonstrates superior purification performance on classification and style mimicry tasks, exposing critical vulnerabilities in black-box data protection. We suggest that practitioners implement multi-level countermeasures to mitigate such risks.
Poster	BridgeVLA: Input-Output Alignment for Efficient 3D Manipulation Learning with Vision-Language Models	https://neurips.cc//virtual/2025/poster/116823	Peiyan Li, Yixiang Chen, Hongtao Wu, Xiao Ma, Xiangnan Wu, Yan Huang, Liang Wang, Tao Kong, Tieniu Tan	Recently, leveraging pre-trained vision-language models (VLMs) for building vision-language-action (VLA) models has emerged as a promising approach to effective robot manipulation learning. However, only few methods incorporate 3D signals into VLMs for action prediction, and they do not fully leverage the spatial structure inherent in 3D data, leading to low sample efficiency. In this paper, we introduce BridgeVLA, a novel 3D VLA model that (1) projects 3D inputs to multiple 2D images, ensuring input alignment with the VLM backbone, and (2) utilizes 2D heatmaps for action prediction, unifying the input and output spaces within a consistent 2D image space.In addition, we propose a scalable pre-training method that equips the VLM backbone with the capability to predict 2D heatmaps before downstream policy learning. Extensive experiments show the proposed method is able to learn 3D manipulation efficiently and effectively.BridgeVLA surpasses the state-of-the-art baseline method in RLBench, achieving a significant higher success rate (88.2% vs 81.4%), and in COLOSSEUM, demonstrating a substantially lower success rate drop (3.6% vs 15.6%). In real-robot experiments, BridgeVLA outperforms the state-of-the-art baseline method by 32% on average, and is able to generalize robustly in multiple out-of-distribution settings, including visual disturbance and unseen instructions. Remarkably, it is able to achieve a success rate of 96.8% on 10+ tasks with only 3 trajectories per task, highlighting its extraordinary sample efficiency. Videos can be found in https://anonymous1219-create.github.io/BridgeVLA_Web/.
Poster	Bridging Arbitrary and Tree Metrics via Differentiable Gromov Hyperbolicity	https://neurips.cc//virtual/2025/poster/115796	Pierre Houédry, Nicolas Courty, Florestan Martin-Baillon, Laetitia Chapel, Titouan Vayer	Trees and the associated shortest-path tree metrics provide a powerful framework for representing hierarchical and combinatorial structures in data. Given an arbitrary metric space, its deviation from a tree metric can be quantified by Gromov’s $\delta$-hyperbolicity. Nonetheless, designing algorithms that bridge an arbitrary metric to its closest tree metric is still a vivid subject of interest, as most common approaches are either heuristical and lack guarantees, or perform moderately well. In this work, we introduce a novel differentiable optimization framework, coined \ourmethod, that solves this problem. Our method leverages a smooth surrogate for Gromov’s $\delta$-hyperbolicity which enables a gradient-based optimization, with a tractable complexity. The corresponding optimization procedure is derived from a problem with better worst case guarantees than existing bounds, and is justified statistically. Experiments on synthetic and real-world datasets demonstrate that our method consistently achieves state-of-the-art distortion.
Poster	Bridging Brains and Concepts: Interpretable Visual Decoding from fMRI with Semantic Bottlenecks	https://neurips.cc//virtual/2025/poster/118670	Sara Cammarota, Matteo Ferrante, Nicola Toschi	Decoding of visual stimuli from noninvasive neuroimaging techniques such as functional magnetic resonance has advanced rapidly in the last years; yet, most high-performing brain decoding models rely on complicated, non intelligible latent spaces. In this study we present an interpretable brain decoding framework that inserts a semantic bottleneck into BrainDiffuser, a well established, simple and linear decoding pipeline. We firstly produce a $214-\text{dimensional}$ binary interpretable space $\mathcal{L}$ for images, in which each dimension answers to a specific question about the image (e.g., "Is there a person?", "Is it outdoors?"). A first ridge regression maps voxel activity to this semantic space. Because this mapping is linear, its weight matrix can be visualized as maps of voxel importance for each dimension of $\mathcal{L}$, revealing which cortical regions influence mostly each semantic dimension. A second regression then transforms these concept vectors into CLIP embeddings required to produce the final decoded image, conditioning the BrainDiffuser model. We found that voxel-wise weight maps for individual questions are highly consistent with canonical category-selective regions in the visual cortex (face, bodies, places, words), simultaneously revealing that activation distributions, not merely location, bear semantic meaning in the brain. Visual brain decoding performances are only slightly lower compared to the original BrainDiffuser metrics (e.g., the CLIP similarity is decreased by $\leq 4$% for the four subjects), yet offering substantial gains in interpretability and neuroscientific insights. These results show that our interpretable brain decoding pipeline enables voxel-level analysis of semantic representations in the human brain without sacrificing decoding accuracy.
Poster	Bridging Critical Gaps in Convergent Learning: How Representational Alignment Evolves Across Layers, Training, and Distribution Shifts	https://neurips.cc//virtual/2025/poster/117325	Chaitanya Kapoor, Sudhanshu Srivastava, Meenakshi Khosla	Understanding convergent learning — the degree to which independently trained neural systems — whether multiple artificial networks or brains and models — arrive at similar internal representations — is crucial for both neuroscience and AI. Yet, the literature remains narrow in scope—typically examining just a handful of models with one dataset, relying on one alignment metric, and evaluating networks at a single post-training checkpoint. We present a large-scale audit of convergent learning, spanning dozens of vision models and thousands of layer-pair comparisons, to close these long-standing gaps. First, we pit three alignment families against one another---linear regression (affine-invariant), orthogonal Procrustes (rotation-/reflection-invariant), and permutation/soft-matching (unit-order-invariant). We find that orthogonal transformations align representations nearly as effectively as more flexible linear ones, and although permutation scores are lower, they significantly exceed chance, indicating a privileged representational basis. Tracking convergence throughout training further shows that nearly all eventual alignment crystallizes within the first epoch---well before accuracy plateaus---indicating it is largely driven by shared input statistics and architectural biases, not by the final task solution. Finally, when models are challenged with a battery of out-of-distribution images, early layers remain tightly aligned, whereas deeper layers diverge in proportion to the distribution shift. These findings fill critical gaps in our understanding of representational convergence, with implications for neuroscience and AI.
Poster	Bridging Crypto with ML-based Solvers: the SAT Formulation and Benchmarks	https://neurips.cc//virtual/2025/poster/121790	Xinhao Zheng, Xinhao Song, Bolin Qiu, Yang Li, Zhongteng Gui, Junchi Yan	The Boolean Satisfiability Problem (SAT) plays a crucial role in cryptanalysis, enabling tasks like key recovery and distinguisher construction. Conflict-Driven Clause Learning (CDCL) has emerged as the dominant paradigm in modern SAT solving, and machine learning has been increasingly integrated with CDCL-based SAT solvers to tackle complex cryptographic problems. However, the lack of a unified evaluation framework, inconsistent input formats, and varying modeling approaches hinder fair comparison. Besides, cryptographic SAT instances also differ structurally from standard SAT problems, and the absence of standardized datasets further complicates evaluation. To address these issues, we introduce SAT4CryptoBench, the first comprehensive benchmark for assessing machine learning–based solvers in cryptanalysis. SAT4CryptoBench provides diverse SAT datasets in both Arithmetic Normal Form (ANF) and Conjunctive Normal Form (CNF), spanning various algorithms, rounds, and key sizes. Our framework evaluates three levels of machine learning integration: standalone distinguishers for instance classification, heuristic enhancement for guiding solving strategies, and hyperparameter optimization for adapting to specific problem distributions. Experiments demonstrate that ANF-based networks consistently achieve superior performance over CNF-based networks in learning cryptographic features. Nonetheless, current ML techniques struggle to generalize across algorithms and instance sizes, with computational overhead potentially offsetting benefits on simpler cases. Despite this, ML-driven optimization strategies notably improve solver efficiency on cryptographic SAT instances. Besides, we propose BASIN, a bitwise solver taking plaintext-ciphertext bitstrings as inputs. Crucially, its superior performance on high-round problems highlights the importance of input modeling and the advantage of direct input representations for complex cryptographic structures.
Poster	Bridging Equivariant GNNs and Spherical CNNs for Structured Physical Domains	https://neurips.cc//virtual/2025/poster/115117	Colin Kohler, Purvik Patel, Nathan Vaska, Justin Goodwin, Matthew Jones, Robert Platt, Rajmonda Caceres, Robin Walters	Many modeling tasks from disparate domains can be framed the same way, computing spherical signals from geometric inputs, for example, computing the radar response of different objects or navigating through an environment. This paper introduces G2Sphere, a general method for mapping object geometries to spherical signals. G2Sphere operates entirely in Fourier space, encoding geometric structure into latent Fourier features using equivariant neural networks and outputting the Fourier coefficients of the continuous target signal, which can be evaluated at any resolution. By utilizing a hybrid GNN-spherical CNN architecture, our method achieves much higher frequency output signal than comparable equivariant GNNs and avoids hand-engineered geometry features used previously by purely spherical methods. We perform experiments on various challenging domains including radar response modeling, aerodynamic drag prediction, and policy learning for manipulation and navigation. We find that G2Sphere outperforms competitive baselines in terms of accuracy and inference time, and we demonstrate that equivariance and Fourier features lead to improved sample efficiency and generalization.
Poster	Bridging Expressivity and Scalability with Adaptive Unitary SSMs	https://neurips.cc//virtual/2025/poster/115719	Arjun Karuvally, Franz Nowak, Andy Keller, Carmen Amo Alonso, Terrence Sejnowski, Hava Siegelmann	Recent work has revealed that state space models (SSMs), while efficient for long-sequence processing, are fundamentally limited in their ability to represent formal languages—particularly due to time-invariant and real-valued recurrence structures. In this work, we draw inspiration from adaptive and structured dynamics observed in biological neural systems and introduce the Adaptive Unitary State Space Model (AUSSM): a novel class of SSMs that leverages skew-symmetric, input-dependent recurrence to achieve unitary evolution and high expressive power. Using algebraic automata theory, we prove that AUSSM can perform modulo counting and simulate solvable group automata at finite precision, enabling SSMs to model a broad class of regular languages out of reach for other SSM architectures. To overcome the practical inefficiencies of adaptive recurrence, we develop a separable convolution formulation and a CUDA implementation that enables scalable parallel training. Empirically, we show that AUSSM and its hybrid variant—interleaved with Mamba—outperform prior SSMs on formal algorithmic tasks such as parity and modular arithmetic, and achieve competent performance on real-world long time-series classification benchmarks. Our results demonstrate that adaptive unitary recurrence provides a powerful and efficient inductive bias for both symbolic and continuous sequence modeling.
Poster	Bridging Gaze and VLMs through Attention Regularization for Egocentric Understanding	https://neurips.cc//virtual/2025/poster/120280	Anupam Pani, Yanchao Yang	Eye gaze offers valuable cues about attention, short-term intent, and future actions, making it a powerful signal for modeling egocentric behavior. In this work, we propose a gaze-regularized framework that enhances VLMs for two key egocentric understanding tasks: fine-grained future event prediction and current activity understanding.Unlike prior approaches that rely solely on visual inputs or use gaze as an auxiliary input signal , our method uses gaze only during training. We introduce a gaze-regularized attention mechanism that aligns model focus with human visual gaze. This design is flexible and modular, allowing it to generalize across multiple VLM architectures that utilize attention.Experimental results show that our approach improves semantic prediction scores by up to 11 $\%$ for future event prediction and around 7 $\%$ for current activity understanding, compared to the corresponding baseline models trained without gaze regularization. These results highlight the value of gaze-guided training in improving the accuracy and robustness of egocentric VLMs. Overall, this work establishes a foundation for using human gaze to enhance the predictive capabilities of VLMs in real-world scenarios like assistive robots and human-machine collaboration.
Poster	Bridging Human and LLM Judgments: Calibration, Alignment, and Bias Detection	https://neurips.cc//virtual/2025/poster/117193	Felipe Maia Polo, Xinhe Wang, Mikhail Yurochkin, Gongjun Xu, Moulinath Banerjee, Yuekai Sun	LLM-as-a-judge (LLMJ) has become popular for scalably evaluating language model outputs on open-ended user queries. However, LLM judges do not always align with human annotators, exhibiting systematic and undesired differences, e.g., biases toward certain writing styles. In this paper, we introduce the first unified statistical framework that jointly models human and LLM judge ratings under both absolute and relative (e.g., pairwise comparisons) evaluation paradigms. We posit the existence of a latent human preference score that is shared between humans and LLM judges for each prompt–response pair, and allow LLM judgments to deviate systematically via a linear transformation of a covariate vector, which encodes potential sources of LLM judge biases. Our core contribution is a statistical model that links LLM-assigned scores to the latent human signal and a set of biasing covariates, for which we propose a fitting algorithm and establish asymptotic normality for resulting estimators, enabling formal hypothesis tests of human-LLM discrepancies. Empirically, our model (i) improves the probability calibration of LLM judgments, (ii) aligns LLM ratings more closely with human judgments, and (iii) quantifies and tests systematic LLM biases across evaluation scenarios. We verify the efficacy and provide insightful results using four LLM judges and queries from BigGen Bench and Chatbot Arena.
Poster	Bridging Scales: Spectral Theory Reveals How Local Connectivity Rules Sculpt Global Neural Dynamics in Spatially Extended Networks	https://neurips.cc//virtual/2025/poster/120113	Yuhan Huang, Keren Gao, Dongping Yang, Sen Song, Guozhang Chen	The brain's diverse spatiotemporal activity patterns are fundamental to cognition and consciousness, yet how these macroscopic dynamics emerge from microscopic neural circuitry remains a critical challenge. We address this by developing a spatially extended neural network model integrated with a spectral theory of its connectivity matrix. Our theory quantitatively demonstrates how local structural parameters, such as E/I neuron projection ranges, connection strengths and density determine distinct features of the eigenvalue spectrum, specifically outlier eigenvalues and a bulk disk. These spectral signatures, in turn, precisely predict the network's emergent global dynamical regime, encompassing asynchronous states, synchronous states, oscillations, localized activity bumps, traveling waves, and chaos. Motivated by observations of shifting cortical dynamics in mice across arousal states, our framework not only explains this repertoire of behaviors but also offers a principled approach to inferring underlying effective connectivity changes from macroscopic brain activity. By mechanistically linking neural structure to dynamics, this work provides a powerful tool for understanding brain function and paves they way for identifying potential biomarkers for neurological disorders.
Poster	Bridging Sign and Spoken Languages: Pseudo Gloss Generation for Sign Language Translation	https://neurips.cc//virtual/2025/poster/115986	Jianyuan Guo, Peike Li, Trevor Cohn	Sign Language Translation (SLT) aims to map sign language videos to spoken language text. A common approach relies on gloss annotations as an intermediate representation, decomposing SLT into two sub-tasks: video-to-gloss recognition and gloss-to-text translation. While effective, this paradigm depends on expert-annotated gloss labels, which are costly and rarely available in existing datasets, limiting its scalability. To address this challenge, we propose a gloss-free pseudo gloss generation framework that eliminates the need for human-annotated glosses while preserving the structured intermediate representation.Specifically, we prompt a Large Language Model (LLM) with a few example text-gloss pairs using in-context learning to produce draft sign glosses from spoken language text. To enhance the correspondence between LLM-generated pseudo glosses and the sign sequences in video, we correct the ordering in the pseudo glosses for better alignment via a weakly supervised learning process.This reordering facilitates the incorporation of auxiliary alignment objectives, and allows for the use of efficient supervision via a Connectionist Temporal Classification (CTC) loss.We train our SLT model—consisting of a vision encoder and a translator—through a three-stage pipeline, which progressively narrows the modality gap between sign language and spoken language.Despite its simplicity, our approach outperforms previous state-of-the-art gloss-free frameworks on two SLT benchmarks and achieves competitive results compared to gloss-based methods.
Poster	Bridging Symmetry and Robustness: On the Role of Equivariance in Enhancing Adversarial Robustness	https://neurips.cc//virtual/2025/poster/115259	Longwei Wang, Chaowei Zhang, Ifrat Ikhtear Uddin, Prof. KC Santosh (PhD), Xiao Qin, Yang Zhou	Adversarial examples reveal critical vulnerabilities in deep neural networks by exploiting their sensitivity to imperceptible input perturbations. While adversarial training remains the predominant defense strategy, it often incurs significant computational cost and may compromise clean-data accuracy. In this work, we investigate an architectural approach to adversarial robustness by embedding group-equivariant convolutions—specifically, rotation- and scale-equivariant layers—into standard convolutional neural networks (CNNs). These layers encode symmetry priors that align model behavior with structured transformations in the input space, promoting smoother decision boundaries and greater resilience to adversarial attacks. We propose and evaluate two symmetry-aware architectures: a parallel design that processes standard and equivariant features independently before fusion, and a cascaded design that applies equivariant operations sequentially. Theoretically, we demonstrate that such models reduce hypothesis space complexity, regularize gradients, and yield tighter certified robustness bounds under the CLEVER (Cross Lipschitz Extreme Value for nEtwork Robustness) framework. Empirically, our models consistently improve adversarial robustness and generalization across CIFAR-10, CIFAR-100, and CIFAR-10C under both FGSM and PGD attacks, without requiring adversarial training. These findings underscore the potential of symmetry-enforcing architectures as efficient and principled alternatives to data augmentation-based defenses.
Poster	Bridging the Expressivity Gap: Provably Tractable SHAP Explanations for Tensor Networks	https://neurips.cc//virtual/2025/poster/119025	Reda Marzouk, Shahaf Bassan, Guy Katz	Although Shapley additive explanations (SHAP) can be computed in polynomial time for simple models like decision trees, they unfortunately become NP-hard to compute for more expressive black-box models like neural networks - where generating explanations is often most critical. In this work, we analyze the problem of computing SHAP explanations for Tensor Networks (TNs), a broader and more expressive class of models than those for which current exact SHAP algorithms are known to hold, and which is widely used for neural network abstraction and compression. First, we introduce a general framework for computing provably exact SHAP explanations for general TNs with arbitrary structures. Interestingly, we show that, when TNs are restricted to a Tensor Train (TT) structure, SHAP computation can be performed in poly-logarithmic time using parallel computation. Thanks to the expressiveness power of TTs, this complexity result can be generalized to many other popular ML models such as decision trees, tree ensembles, linear models, and linear RNNs, therefore tightening previously reported complexity results for these families of models. Finally, by leveraging reductions of binarized neural networks to Tensor Network representations, we demonstrate that SHAP computation can become efficiently tractable when the network’s width is fixed, while it remains computationally hard even with constant depth. This highlights an important insight: for this class of models, width - rather than depth - emerges as the primary computational bottleneck in SHAP computation.
Poster	Bridging the Gap Between Cross-Domain Theory and Practical Application: A Case Study on Molecular Dissolution	https://neurips.cc//virtual/2025/poster/117544	Sihan Wang, Wenjie Du, Qing Zhu, Yang Wang	Artificial intelligence (AI) has played a transformative role in chemical research, greatly facilitating the prediction of small molecule properties, simulation of catalytic processes, and material design. These advances are driven by increases in computing power, open source machine learning frameworks, and extensive chemical datasets. However, a persistent challenge is the limited amount of high-quality real-world data, while models calculated based on large amounts of theoretical data are often costly and difficult to deploy, which hinders the applicability of AI models in real-world scenarios. In this study, we enhance the prediction of solute-solvent properties by proposing a novel sample selection method: the iterative core subset extraction (CSIE) framework. CSIE iteratively updates the core sample subset based on information gain to remove redundant features in theoretical data and optimize the performance of the model on real chemical datasets. Furthermore, we introduce an asymmetric molecular interaction graph neural network (AMGNN) that combines positional information and bidirectional edge connections to simulate real-world chemical reaction scenarios to better capture solute-solvent interactions. Experimental results show that our method can accurately extract the core subset and improve the prediction accuracy.
Poster	Bridging Theory and Practice in Link Representation with Graph Neural Networks	https://neurips.cc//virtual/2025/poster/117590	Veronica Lachi, Francesco Ferrini, Antonio Longa, Bruno Lepri, Andrea Passerini, Manfred Jaeger	Graph Neural Networks (GNNs) are widely used to compute representations of node pairs for downstream tasks such as link prediction. Yet, theoretical understanding of their expressive power has focused almost entirely on graph-level representations. In this work, we shift the focus to links and provide the first comprehensive study of GNN expressiveness in link representation. We introduce a unifying framework, the $k_\phi$-$k_\rho$-$m$ framework, that subsumes existing message-passing link models and enables formal expressiveness comparisons. Using this framework, we derive a hierarchy of state-of-the-art methods and offer theoretical tools to analyze future architectures. To complement our analysis, we propose a synthetic evaluation protocol comprising the first benchmark specifically designed to assess link-level expressiveness. Finally, we ask: does expressiveness matter in practice? We use a graph symmetry metric that quantifies the difficulty of distinguishing links and show that while expressive models may underperform on standard benchmarks, they significantly outperform simpler ones as symmetry increases, highlighting the need for dataset-aware model selection.
Poster	Bridging Time and Linguistics: LLMs as Time Series Analyzer through Symbolization and Segmentation	https://neurips.cc//virtual/2025/poster/116138	Jianyang Qin, Chaoyang Li, Jinhao Cui, Lingzhi Wang, Zhao Liu, Qing Liao	Recent studies reveal that Large Language Models (LLMs) exhibit strong sequential reasoning capabilities, allowing them to replace specialized time-series models and serve as foundation models for complex time-series analysis. To activate the capabilities of LLMs for time-series tasks, numerous studies have attempted to bridge the gap between time series and linguistics by aligning textual representations with time-series patterns. However, it is a non-trivial endeavor to losslessly capture the infinite time-domain variability using natural language, leading to suboptimal alignment performance. Beyond representation, contextual differences, where semantics in time series are conveyed by consecutive points, unlike in text by individual tokens, are often overlooked by existing methods. To address these, we propose S$^2$TS-LLM, a simple yet effective framework to repurpose LLMs for universal time series analysis through the following two main paradigms: (i) a spectral symbolization paradigm transforms time series into frequency-domain representations characterized by a fixed number of components and prominent amplitudes, which enables a limited set of symbols to effectively abstract key frequency features; (ii) a contextual segmentation paradigm partitions the sequence into blocks based on temporal patterns and reassigns positional encodings accordingly, thereby mitigating the structural mismatch between time series and natural language. Together, these paradigms bootstrap the LLMs' perception of temporal patterns and structures, effectively bridging time series and linguistics. Extensive experiments show that S$^2$TS-LLM can serve as a powerful time series analyzer, outperforming state-of-the-art methods across time series tasks.
Poster	Briding the gap to real-world language-grounded visual concept learning	https://neurips.cc//virtual/2025/poster/118837	whie jung, Semin Kim, Junee Kim, Seunghoon Hong	Human intelligence effortlessly interprets visual scenes along a rich spectrum of semantic dimensions. However, existing approaches for language-grounded visual concept learning are limited to a few predefined primitive axes such as color and shape, and are explored in synthetic datasets.In this work, we propose a scalable framework that adaptively identifies image-related concept axes and grounds visual concepts along these axes in real-world scenes. Leveraging a pretrained vision-language model with our simple universal prompting strategy, our framework identifies a diverse image-related axes without requiring any prior knowledge. Our universal concept encoder then adaptively binds visual features to the discovered axes without introducing additional model parameters per concept. To ground visual concepts along discovered axes, we maximize the compositional consistency of concept representations, which ensures each axis to be independently manipulated without affecting other axes.We demonstrate the effectiveness of our framework on CelebA-HQ and AFHQ datasets, achieving superior editing capabilities across diverse concepts and strong compositional generalization compared to existing visual concept learning method and text-based editing methods.
Poster	Bringing SAM to new heights: leveraging elevation data for tree crown segmentation from drone imagery	https://neurips.cc//virtual/2025/poster/120185	Mélisande Teng, Arthur Ouaknine, Etienne Laliberté, Yoshua Bengio, David Rolnick, Hugo Larochelle	Information on trees at the individual level is crucial for monitoring forest ecosystems and planning forest management. Current monitoring methods involve ground measurements, requiring extensive cost, time and labour. Advances in drone remote sensing and computer vision offer great potential for mapping individual trees from aerial imagery at broad-scale. Large pre-trained vision models, such as the Segment Anything Model (SAM), represent a particularly compelling choice given limited labeled data. In this work, we compare methods leveraging SAM for the task of automatic tree crown instance segmentation in high resolution drone imagery in three use cases: 1) boreal plantations, 2) temperate forests, and 3) tropical forests. We also look into integrating elevation data into models, in the form of Digital Surface Model (DSM) information, which can readily be obtained at no additional cost from RGB drone imagery. We present BalSAM, a model leveraging SAM and DSM information, which shows potential over other methods, particularly in the context of plantations. We find that methods using SAM out-of-the-box do not outperform a custom Mask R-CNN, even with well-designed prompts. However, efficiently tuning SAM further and integrating DSM information are both promising avenues for tree crown instance segmentation models.
Poster	Broken Tokens? Your Language Model can Secretly Handle Non-Canonical Tokenizations	https://neurips.cc//virtual/2025/poster/117561	Brian Zheng, Alisa Liu, Orevaoghene Ahia, Jonathan Hayase, Yejin Choi, Noah Smith	Modern tokenizers employ deterministic algorithms to map text into a single ``canonical" token sequence, yet the same string can be encoded as many non-canonical tokenizations using the language model vocabulary, including tokenizing by character. In this paper, we investigate the robustness of LMs to input encoded with non-canonical tokenizations entirely unseen during training. Surprisingly, when evaluated across 20 benchmarks, we find that instruction-tuned models retain up to 93.4\% of their original performance when given a randomly sampled tokenization, and 90.8\% with character-level tokenization. We find that overall stronger models tend to be more robust, and that robustness diminishes as the tokenization departs farther from the canonical form. Motivated by these results, we identify settings where non-canonical tokenization schemes can \textit{improve} performance, finding that character‑level segmentation improves string manipulation and code understanding tasks by up to 15\%, and right‑aligned digit grouping enhances large‑number arithmetic by over 33\%. Finally, we investigate the source of this robustness, finding that it arises in the instruction-tuning phase. We provide evidence that both base and post-trained models grasp the semantics of non-canonical tokenizations (perceiving them as containing misspellings). However, base models try to mimic the imagined mistakes and degenerate into nonsensical output, while post-trained models are committed to fluent responses. Overall, our findings suggest that models are less committed to their tokenizer than previously believed, and highlight the promise of intervening on tokenization at inference time to boost language model performance.
Poster	BTL-UI: Blink-Think-Link Reasoning Model for GUI Agent	https://neurips.cc//virtual/2025/poster/119419	Shaojie Zhang, Ruoceng Zhang, Pei Fu, Shaokang Wang, Jiahui Yang, Xin Du, ShiqiCui, Bin Qin, Ying Huang, Zhenbo Luo, Jian Luan	In the field of AI-driven human-GUI interaction automation, while rapid advances in multimodal large language models and reinforcement fine-tuning techniques have yielded remarkable progress, a fundamental challenge persists: their interaction logic significantly deviates from natural human-GUI communication patterns. To fill this gap, we propose ``Blink-Think-Link'' (BTL), a brain-inspired framework for human-GUI interaction that mimics the human cognitive process between users and graphical interfaces. The system decomposes interactions into three biologically plausible phases: (1) \textbf{Blink} - rapid detection and attention to relevant screen areas, analogous to saccadic eye movements; (2) \textbf{Think} - higher-level reasoning and decision-making, mirroring cognitive planning; and (3) \textbf{Link} - generation of executable commands for precise motor control, emulating human action selection mechanisms. Additionally, we introduce two key technical innovations for BTL framework:(1) Blink Data Generation - an automated annotation pipeline specifically optimized for blink data, and(2) {BTL Reward – the first rule-based reward mechanism that enables reinforcement learning driven by both process and outcome.}Building upon this framework, we develop a GUI agent model named BTL-UI, which demonstrates consistent state-of-the-art performance across both static GUI understanding and dynamic interaction tasks in comprehensive benchmarks. These results provide conclusive empirical validation of the framework's efficacy in developing advanced GUI Agents. We will soon release the relevant data and models.
Poster	Bubbleformer: Forecasting Boiling with Transformers	https://neurips.cc//virtual/2025/poster/121854	Sheikh Md Shakeel Hassan, Xianwei Zou, Akash Dhruv, Aparna Chandramowlishwaran	Modeling boiling--an inherently chaotic, multiphase process central to energy and thermal systems--remains a significant challenge for neural PDE surrogates. Existing models rely on simulation inputs (e.g., bubble positions), limiting their ability to forecast future states autonomously. We introduce Bubbleformer, a transformer-based spatiotemporal model that forecasts stable and long-range boiling dynamics including nucleation, interface evolution, and heat transfer without dependence on simulation data during inference. Bubbleformer combines axial attention with physics-informed architectural modifications to mitigate spectral bias and conditions on thermophysical parameters to generalize across fluids, geometries, and operating conditions. To evaluate models of these chaotic systems, we introduce physically grounded metrics. We also release BubbleML 2.0, a high-fidelity dataset spanning diverse working fluids (cryogens, refrigerants, and dielectrics), boiling configurations (pool boiling and flow boiling), distinct flow boiling regimes (bubbly, slug, and annular), and operational and boundary conditions. Bubbleformer sets new benchmarks in both prediction and forecasting tasks of two-phase boiling flows.
Poster	BundleFlow: Deep Menus for Combinatorial Auctions by Diffusion-Based Optimization	https://neurips.cc//virtual/2025/poster/116578	Tonghan Wang, Yanchen Jiang, David Parkes	Differentiable economics---the use of deep learning for auction design---has driven progress in multi-item auction design with additive or unit-demand valuations. However, progress in combinatorial auctions (CAs), even for the simplest and yet important single bidder case, remains limited due to the exponential growth of the bundle space with the number of items. We address this challenge by introducing a deep menu-based mechanism, which is, to our knowledge, the first dominant-strategy incentive compatible (DSIC) and revenue-optimizing single-bidder CA. Our idea is to generate a bundle distribution through an ordinary differential equation (ODE) applied to a tractable initial distribution. Our method learns suitable ODE-based transforms, one for each menu element, to optimize expected revenue. Our method achieves 1.11$-$2.23$\times$ higher revenue than baselines on standard CA testbeds and scales up to 150 items. Relative to menu-learning baselines that we introduce, our method also reduces training iterations by 3.6$-$9.5$\times$ and cuts training time by about 80\% in settings with 50 and 100 items.
Poster	BurstDeflicker: A Benchmark Dataset for Flicker Removal in Dynamic Scenes	https://neurips.cc//virtual/2025/poster/121576	Lishen Qu, Zhihao Liu, Shihao Zhou, LUO YAQI, Jie Liang, Hui Zeng, Lei Zhang, Jufeng Yang	Flicker artifacts in short-exposure images are caused by the interplay between the row-wise exposure mechanism of rolling shutter cameras and the temporal intensity variations of alternating current (AC)-powered lighting. These artifacts typically appear as uneven brightness distribution across the image, forming noticeable dark bands. Beyond compromising image quality, this structured noise also affects high-level tasks, such as object detection and tracking, where reliable lighting is crucial. Despite the prevalence of flicker, the lack of a large-scale, realistic dataset has been a significant barrier to advancing research in flicker removal. To address this issue, we present BurstDeflicker, a robust and scalable benchmark constructed using three complementary data acquisition strategies. First, we develop a Retinex-based synthesis pipeline that redefines the goal of flicker removal and enables controllable manipulation of key flicker-related attributes (e.g., intensity, area, and frequency), thereby facilitating the generation of diverse flicker patterns. Second, we capture 4,000 real-world flicker images from different scenes, which help the model better understand the spatial and temporal characteristics of real flicker artifacts and generalize more effectively to wild scenarios. Finally, due to the non-repeatable nature of dynamic scenes, we propose a green-screen method to incorporate motion into image pairs while preserving real flicker degradation. Comprehensive experiments demonstrate the effectiveness of our dataset and its potential to advance research in flicker removal. The code and dataset are available in the supplementary materials.
Poster	C${}^2$Prompt: Class-aware Client Knowledge Interaction for Federated Continual Learning	https://neurips.cc//virtual/2025/poster/115966	Kunlun Xu, Yibo Feng, Jiangmeng Li, Yongsheng Qi, Jiahuan Zhou	Federated continual learning (FCL) tackles scenarios of learning from continuously emerging task data across distributed clients, where the key challenge lies in addressing both temporal forgetting over time and spatial forgetting simultaneously. Recently, prompt-based FCL methods have shown advanced performance through task-wise prompt communication.In this study, we underscore that the existing prompt-based FCL methods are prone to class-wise knowledge coherence between prompts across clients. The class-wise knowledge coherence includes two aspects: (1) intra-class distribution gap across clients, which degrades the learned semantics across prompts, (2) inter-prompt class-wise relevance, which highlights cross-class knowledge confusion. During prompt communication, insufficient class-wise coherence exacerbates knowledge conflicts among new prompts and induces interference with old prompts, intensifying both spatial and temporal forgetting. To address these issues, we propose a novel Class-aware Client Knowledge Interaction (C${}^2$Prompt) method that explicitly enhances class-wise knowledge coherence during prompt communication. Specifically, a local class distribution compensation mechanism (LCDC) is introduced to reduce intra-class distribution disparities across clients, thereby reinforcing intra-class knowledge consistency. Additionally, a class-aware prompt aggregation scheme (CPA) is designed to alleviate inter-class knowledge confusion by selectively strengthening class-relevant knowledge aggregation. Extensive experiments on multiple FCL benchmarks demonstrate that C${}^2$Prompt achieves state-of-the-art performance. Our code will be released.
Poster	C3Po: Cross-View Cross-Modality Correspondence by Pointmap Prediction	https://neurips.cc//virtual/2025/poster/121636	Kuan Wei Huang, Brandon Li, Bharath Hariharan, Noah Snavely	Geometric models like DUSt3R have shown great advances in understanding the geometry of a scene from pairs of photos. However, they fail when the inputs are from vastly different viewpoints (e.g., aerial vs. ground) or modalities (e.g., photos vs. abstract drawings) due to the vast differences in viewpoint or style compared to what was observed during training. This paper addresses a challenging version of this problem: predicting correspondences between ground-level photos and floor plans. Current datasets for joint photo--floor plan reasoning are limited, either lacking in varying modalities (VIGOR) or lacking in correspondences (WAFFLE). To address these limitations, we introduce a new dataset, C3, created by first reconstructing a number of scenes in 3D from Internet photo collections via structure from motion, then manually registering the reconstructions to floor plans gathered from the Internet, from which we can derive correspondence between images and floor plans. C3 contains 91K paired floor plans and photos across 574 scenes with 155M pixel-level correspondences. We find that state-of-the-art correspondence models struggle on this task. By training on our new data, we can improve on the best performing method by 34\% in RMSE. However, we also identify open challenges in cross-modal geometric reasoning that our dataset aims to help address.
Poster	C3PO: Optimized Large Language Model Cascades with Probabilistic Cost Constraints for Reasoning	https://neurips.cc//virtual/2025/poster/116958	Antonios Valkanas, Soumyasundar Pal, Pavel Rumiantsev, Yingxue Zhang, Mark Coates	Large language models (LLMs) have achieved impressive results on complex reasoning tasks, but their high inference cost remains a major barrier to real-world deployment. A promising solution is to use cascaded inference, where small, cheap models handle easy queries, and only the hardest examples are escalated to more powerful models. However, existing cascade methods typically rely on supervised training with labeled data, offer no theoretical generalization guarantees, and provide limited control over test-time computational cost. We introduce C3PO (Cost Controlled Cascaded Prediction Optimization), a self-supervised framework for optimizing LLM cascades under probabilistic cost constraints. By focusing on minimizing regret with respect to the most powerful model, C3PO avoids the need for labeled data by constructing a cascade using only unlabeled model outputs. It leverages conformal prediction to bound the probability that inference cost exceeds a user-specified budget. We provide theoretical guarantees on both cost control and generalization error, and show that our optimization procedure is effective even with small calibration sets. Empirically, C3PO achieves state-of-the-art performance across a diverse set of reasoning benchmarks—including GSM8K, MATH500, and BigBench-Hard—outperforming strong LLM cascading baselines in both accuracy and cost-efficiency. Our results demonstrate that principled, label-free cascade optimization can enable scalable and reliable LLM deployment.
Poster	Cached Token Similarity Is a Strong Prior for Fine-grained Visual Question Answering	https://neurips.cc//virtual/2025/poster/119637	Liangyu Zhong, Fabio Rosenthal, Joachim Sicking, Fabian Hüger, Thorsten Bagdonat, Hanno Gottschalk, Leo Schwinn	While Multimodal Large Language Models (MLLMs) offer great perception and reasoning capabilities for image-text input, fine-grained Visual Question Answering (VQA) focusing on small details still remains a challenge. Although visual cropping techniques seem promising, recent approaches have several limitations: the need for task-specific fine-tuning, low efficiency due to uninformed exhaustive search, or incompatibility with efficient attention implementations. We address these shortcomings by proposing a training-free visual cropping method, dubbed FOCUS, that leverages MLLM-internal representations to guide the search for the most relevant image region. This is accomplished in four steps: first, we identify the target object(s) in the prompt; second, we compute an object relevance map using the key-value (KV) cache; third, we propose and rank relevant image regions based on the map; and finally, we perform the fine-grained VQA task using the top-ranked region. As a result of this informed search strategy, our method achieves strong performance across four fine-grained VQA datasets and two types of MLLM. It outperforms three existing visual cropping methods in both accuracy and efficiency, and matches the best-performing baseline, ZoomEye, with 3 - 6.5 x higher efficiency. Finally, we perform an ablation study to assess the impact of key design choices.We plan to release our code upon acceptance.
Poster	CAD-Coder: Text-to-CAD Generation with Chain-of-Thought and Geometric Reward	https://neurips.cc//virtual/2025/poster/118098	Guan Yandong, XiMing Xing, Xilin Wang, Jing Zhang, Dong Xu, Qian Yu	In this work, we introduce CAD-Coder, a novel framework that reformulates text-to-CAD as the generation of CadQuery scripts—a Python-based, parametric CAD language.This representation enables direct geometric validation, a richer modeling vocabulary, and seamless integration with existing LLMs. To further enhance code validity and geometric fidelity, we propose a two-stage learning pipeline: (1) supervised fine-tuning on paired text–CadQuery data, and (2) reinforcement learning with Group Reward Policy Optimization (GRPO), guided by a CAD-specific reward comprising both a geometric reward (Chamfer Distance) and a format reward.We also introduce a chain-of-thought (CoT) planning process to improve model reasoning, and construct a large-scale, high-quality dataset of 110K text–CadQuery–3D model triplets and 1.5K CoT samples via an automated pipeline. Extensive experiments demonstrate that CAD-Coder enables LLMs to generate diverse, valid, and complex CAD models directly from natural language, advancing the state of the art of text-to-CAD generation and geometric reasoning.
Poster	CADGrasp: Learning Contact and Collision Aware General Dexterous Grasping in Cluttered Scenes	https://neurips.cc//virtual/2025/poster/119309	Jiyao Zhang, Zhiyuan Ma, Tianhao Wu, Zeyuan Chen, Hao Dong	Dexterous grasping in cluttered environments presents substantial challenges due to the high degrees of freedom of dexterous hands, occlusion, and potential collisions arising from diverse object geometries and complex layouts. To address these challenges, we propose CADGrasp, a two-stage algorithm for general dexterous grasping using single-view point cloud inputs. In the first stage, we predict a scene-decoupled, contact- and collision-aware representation—sparse IBS—as the optimization target. Sparse IBS compactly encodes the geometric and contact relationships between the dexterous hand and the scene, enabling stable and collision-free dexterous grasp pose optimization. To enhance the prediction of this high-dimensional representation, we introduce an occupancy-diffusion model with voxel-level conditional guidance and force closure score filtering. In the second stage, we develop several energy functions and ranking strategies for optimization based on sparse IBS to generate high-quality dexterous grasp poses. Extensive experiments in both simulated and real-world settings validate the effectiveness of our approach, demonstrating its capability to mitigate collisions while maintaining a high grasp success rate across diverse objects and complex scenes.

Subsets and Splits

No community queries yet

The top public SQL queries from the community will appear here once available.