NeurIPS 2026 论文列表

Unsupervised domain adaptation has become an important paradigm for mitigating distribution shift in time series classification. However, existing time series domain adaptation methods typically align source and target data at a fixed time scale and simply fuse time-frequency features before uniform alignment. Furthermore, they do not consider the impact of noisy samples on distribution alignment. To address these limitations, we propose CSPAN (Time-Frequency Decoupled \textbf{C}ross-\textbf{S}cale \textbf{P}artial Optimal Transport \textbf{A}lignment \textbf{N}etwork) for time series domain adaptation. CSPAN first constructs segment-scale time-frequency candidates, and then uses Segment-Scale Aware Top-k Fusion to select the most suitable segment-scale feature representation. It then performs cross-domain alignment after decoupling time and frequency through Temporal Partial Optimal Transport and Frequency Partial Optimal Transport, where temporal domain alignment is guided by the transferability estimated from a probing transport plan, and frequency domain alignment is guided by the discriminability estimated by an auxiliary frequency classifier. By combining segment-scale representation selection, time-frequency decoupling, and partial distribution alignment, CSPAN enables more flexible cross-domain matching. Extensive experiments on six datasets demonstrate CSPAN’s consistent superiority, achieving an average accuracy improvement of 3.16\% in cross-domain scenarios. Code is available at the anonymous link: \url{https://anonymous.4open.science/r/CSPAN-35FB/}.

Corpus reasoning tasks, which require extracting and integrating information from large collections of documents (e.g. scientific literature, LLM agent traces), range from well-studied tasks such as factoid question answering (e.g. *When was Ralph Lauren founded?*), to more complex, underexplored queries like *What are all the conflicting claims in this literature?*. We observe that on these *complex* tasks, efficient architectures can scale much worse compared than they do on more well-explored tasks, but why does this happen? We develop a unified view---corpus task complexity (CTC)---to systematically distinguish complex tasks from simpler ones in terms of the smaller operations needed to solve them (and how such operations scale with corpus size): for example, we classify tasks which compare all possible document pairs as *more complex* than factoid retrieval. We first investigate new long-context language model (LCLM) training dynamics challenges on these tasks, and propose an effective new mask mixing technique that improves O(N) attention approaches on high CTC tasks, but at larger corpus scale we ultimately find that there is no free lunch: (i) methods that scale efficiently with corpus size **do poorly when CTC is high**, and (ii) **methods that *can* handle these tasks scale quadratically in cost with corpus size**. We identify high-CTC tasks as a long-term open problem and encourage more future work investigating new methods to overcome these limitations.

Switching dynamical systems can model complicated time series data while maintaining interpretability by inferring a finite set of dynamics primitives and explaining different portions of the observed time series with one of these primitives. However, due to the discrete nature of this set, such models struggle to capture smooth, variable-speed transitions, as well as stochastic mixtures of overlapping states (e.g., non-instantaneous state transitions), and the inferred dynamics often display spurious rapid switching on real-world datasets. Here, we propose the Gumbel Dynamical Model (GDM). First, by introducing a continuous relaxation of discrete states and a different noise model defined on the relaxed-discrete state space via the Gumbel distribution, GDM expands the set of available state dynamics, allowing the model to approximate smoother and non-stationary ground-truth dynamics more faithfully. Breaking from established literature, this new class directly links states to observations and does not blur latent dynamics with Gaussian noise. Second, the relaxation makes the model fully differentiable, enabling fast and scalable training with standard gradient descent methods. We validate our approach on standard simulation datasets and highlight its ability to model soft, sticky states and transitions in a stochastic setting. Furthermore, we apply our model to two real-world datasets, demonstrating its ability to infer interpretable states in stochastic time series with multiple dynamics, a setting where traditional methods often fail.

Diffusion models, particularly Diffusion Transformers, achieve strong image and video generation quality but remain expensive at inference time due to repeated denoiser evaluations. Feature caching offers a deployment-friendly acceleration strategy by reusing or predicting intermediate representations without retraining the generator or modifying the sampler. However, existing cache-then-forecast methods often rely on Taylor-style finite-difference extrapolation, which can become unstable over longer cache intervals, and typically predict local module outputs whose errors may accumulate through subsequent denoiser blocks. We propose \textbf{BeziCast}, a training-free diffusion acceleration framework that forecasts output-proximal denoising representations using low-order B\'ezier trajectories. BeziCast estimates B\'ezier control points via Tikhonov-stabilized fitting, providing a smooth, capacity-controlled temporal parameterization that avoids explicit high-order derivative estimation and decouples trajectory capacity from cache-query density. To moderate aggressive extrapolation, we further derive equivalent forecasting weights and introduce a convex-hull-inspired guardrail that detects high-risk predictions and softly pulls them toward a conservative simplex-projected estimate. Extensive evaluations across advanced image and video diffusion models demonstrate the effectiveness of BeziCast. In particular, BeziCast accelerates FLUX.1 by up to \(4.79\times\) and HunyuanVideo by up to \(4.11\times\), while preserving substantially better generation quality than competing acceleration baselines.

How can we ensure that safety oversight models used to detect safety violations in AI systems reliably generalise, and how can we understand the factors that influence their generalisation? In this paper, we formalise PAC-Bayes certification for large language model-based safety oversight and obtain non-vacuous PAC-Bayes guarantees for safety oversight models, even under limited data for safety alignment. Building on compression-based PAC-Bayes bounds, we show that highly compressed PEFT adaptations yield extremely short adaptation description lengths, enabling informative and often tight guarantees that certify both classification risk and predictive uncertainty. We introduce a global-scale quantisation method (LoRA-GT) that reduces adaptor description length while preserving model performance, tightening bounds. Our results show that certifiability is strongly linked to adaptor compressibility, with shorter adaptor description lengths yielding tighter guarantees. Empirically, highly compressed adaptations exhibit minimal degradation in performance while enabling substantially stronger certification, suggesting that certifiable large language model oversight may naturally favour low-complexity safety adaptations. We further show that functional distortion tracks both test risk and bound tightness under compression, providing a practical mechanism for selecting simple, certifiable safety adaptations. Together, these results show that compression-based PAC-Bayes analysis provides a practical framework for understanding and designing reliable safety oversight models.

Inverse optimization (IO) seeks to infer the parameters of a decision-maker's objective from observed context-action data. We study *noiseless IO*, where demonstrations are generated by a ground-truth objective. We provide a high-probability $\mathcal{O}(\frac{d}{T})$ generalization bound for the induced action set, where $d$ is the number of unknown parameters and $T$ is the size of the training dataset. We strengthen these guarantees under additional conditions that ensure uniqueness of the chosen action, bringing our IO guarantees in line with best-arm identification results in the bandit literature. We further show that the $\mathcal{O}(\frac{d}{T})$ rate is tight over all consistent estimators considered here, and extend the result to both instantaneous and cumulative regret. Notably, the resulting regret lower bound matches the corresponding upper bounds in the adversarial setting, indicating that the stochastic IO setting is effectively adversarial for the class of estimators studied here. Finally, we propose a parameter-free algorithm with lower per-iteration complexity than generic solvers. Experiments validate the predicted rates and illustrate the tightness of our bounds.

We study cost-aware cascading bandits, where a learner selects an ordered subset of options, tests them sequentially until the first success, and pays the costs of all tested options. In this problem, regret comes from both testing inefficient options and placing options in a suboptimal order, but existing analyses do not separate these effects for inefficient options and therefore yield an inverse-square dependence on the gap $c_i-\theta_i$. We develop a regret decomposition based on intermediate policies that reorder the remaining suffix and remove inefficient options one position at a time. This allows us to quantify the incremental regret incurred when each option is tested. As a consequence, we show that the regret of CC-UCB admits a problem-dependent bound of $O(\sum_{i:\theta_i/c_i<1}\log T/(c_i-\theta_i))$ up to additive terms and a problem-independent bound of order $\tilde O(L\sqrt{T})$. We further prove that the minimax regret is bounded from below by $\Omega(\sqrt{LT})$ by reducing standard multi-armed bandits to a special case of the model. Finally, we propose CC-UCBv2, which removes the need to specify a positive lower bound on costs and handles zero-cost options by separating empirically zero-cost options from the others. Numerical experiments show the effect of misspecified cost lower bounds and demonstrate that the proposed modification can reduce regret in representative instances involving zero or misspecified costs.

Despite the remarkable progress in Multimodal Large Language Models (MLLMs), the prevailing Chain-of-Thought (CoT) paradigms remain predominantly confined to the linguistic space. While recent advancements have focused on bridging the Perception Gap through high-resolution cropping (e.g., Thinking with Images), they overlook a more fundamental bottleneck: the Reference Gap. The inherent ambiguity of natural language often fails to provide precise, unambiguous pointers to complex spatial layouts, leading to logical collapse in tasks requiring rigorous grounding. In this work, we introduce Thinking with Visual Primitives, a novel reasoning framework that elevates spatial markers (e.g., points and bounding boxes) to "minimal units of thought". By interleaving these visual primitives directly into the thinking process, our model can "point" while it "reasons", effectively grounding its cognitive trajectory in the physical coordinates of the image. Notably, our framework is built on a highly optimized architecture with extreme visual token efficiency. Despite its compact model scale and significantly lower image-token budget, our model achieves frontier-competitive performance on a focused suite of challenging visual QA tasks, matching or exceeding models such as GPT-5.4, Claude-Sonnet-4.6, and Gemini-3-Flash. This demonstrates a path toward more efficient and scalable System-2-like multimodal intelligence.

While Vision-Language Models (VLMs) have shown strong visual reasoning capabilities, their spatial reasoning abilities remain largely constrained to the observed images and text-oriented chain-of-thought. They often struggle to infer unobserved layouts, maintain cross-view consistency, and reason from alternative viewpoints when only limited egocentric observations are available. In this work, we study this problem as thinking with imagination, where a VLM actively acquires imagined visual evidence by interacting with a world simulator during reasoning. We propose Astra, an agentic spatial reasoning framework that empowers VLMs with action-conditioned visual imagination. Specifically, Astra couples Astra-VL, an RL-trained VLM policy, with Astra-WM, a Bagel-based world simulator that generates novel-view observations from context images and natural-language camera motions. To provide reliable imagined evidence, Astra-WM is trained with view consistency tuning to improve pose and content consistency across views. In the RL stage, we propose a world-simulator-in-the-loop two-phase RL curriculum to stabilize tool-use exploration and advance the model's ability to invoke the simulator only when imagined observations improve over direct answering. Experiments demonstrate that both the world simulator and the agentic policy are necessary: Astra-WM improves simulator-augmented Gemini-3-Flash on MMSI-Bench from 45.1 to 49.5, while Astra-VL improves the Qwen3-VL backbone from 29.8 to 38.8 on MMSI-Bench and from 36.8 to 42.7 on MindCube. These results show that imagined observations can provide useful spatial evidence, but effective world-model-augmented reasoning requires learning when, where, and how to imagine.

Recent multimodal large language models (MLLMs) increasingly rely on visual chain-of-thought to perform region-grounded reasoning over images. However, existing approaches ground regions via either textified coordinates—causing modality mismatch and semantic fragmentation—or fixed-granularity patches that both limit precise region selection and often require non-trivial architectural changes. In this paper, we propose Numerical Visual Chain-of-Thought (NV-CoT), a framework that enables MLLMs to reason over images using continuous numerical coordinates. NV-CoT expands the MLLM action space from discrete vocabulary tokens to a continuous Euclidean space, allowing models to directly generate bounding-box coordinates as actions with only minimal architectural modification. The framework supports both supervised fine-tuning and reinforcement learning. In particular, we replace categorical token policies with a Gaussian (or Laplace) policy over coordinates and introduce stochasticity via reparameterized sampling, making NV-CoT fully compatible with GRPO-style policy optimization. Extensive experiments on three benchmarks against eight representative visual reasoning baselines demonstrate that NV-CoT significantly improves localization precision and final answer accuracy, while also accelerating training convergence, validating the effectiveness of continuous-action visual reasoning in MLLMs. The code is provided in the Supplementary Materials.

Recent advancements in world models and unified generative models (UGMs) have achieved unprecedented results in visual perception and synthesis. However, these models primarily rely on surface-level alignment, leaving the capacity for high-level visual reasoning underexplored. True visual generative intelligence demands "Reasoning-to-Generation", an ability to infer latent rules from visual inputs and manifest solutions through precise, logically constrained visual outcomes. We introduce RIG-Bench, a novel comprehensive benchmark that systematically evaluates Reasoning-driven Image Generation (RIG) across four cognitively demanding domains: Concept-based, Transformation-based, Pattern & Structure, and Scenario-based. Featuring 2000 curated samples, RIG-Bench serves as a rigorous stress test for RIG. Our extensive evaluations of state-of-the-art UGMs and image/video generation models reveal a significant reasoning-generation gap, wherein models frequently produce locally plausible but globally illogical outputs. RIG-Bench provides a vital diagnostic framework to guide the development of next-generation, logically grounded world models.

Text and 2D-conditioning interfaces provide weak, ambiguous control over spatial transformations in image editing -- particularly under large object motions and camera changes. Prior work has used 3D primitives such as boxes, but only as loose conditioning signals indicating approximate object location rather than specifying the transformation. We instead use 3D boxes as structured specifications: the user provides the input and output boxes of the edit, casting editing as a well-posed geometry problem. This ``thinking in boxes'' interface, where each box face is color-coded to convey 3D orientation, gives precise control over translation, rotation, scaling, and viewpoint changes in real images while preserving scene and object identity, and recovering previously unseen object regions. To ground transformations in scene appearance, we introduce a depth-aligned planar floor as a global reference frame, shaded with depth-aware cues. Conditioned on this structure, an image generator produces consistent results under large transformations. Trained in two stages -- on synthetic multi-object scenes and a small set of real-world videos from Objectron -- the system generalizes to complex, in-the-wild real images. Our method operates directly on real photographs and substantially outperforms recent state-of-the-art methods on large 3D edits.

Large reasoning models (LRMs) achieve remarkable performance by leveraging reinforcement learning (RL) on reasoning tasks to generate long chain-of-thought (CoT) reasoning. However, this over-optimization often prioritizes compliance, making models vulnerable to harmful prompts. To mitigate this safety degradation, recent approaches rely on external teacher distillation, yet this introduces a \emph{distributional discrepancy} that degrades native reasoning. We formalize safety realignment as a KL projection onto the safe simplex and prove that the student's own safety-filtered distribution is the unique KL-optimal target, while any external teacher incurs an irreducible excess KL penalty. Guided by this analysis, we propose ThinkSafe, a self-generated alignment framework that restores safety without external teachers. Our key insight is that while compliance suppresses safety mechanisms, models often retain latent knowledge to identify harm. ThinkSafe unlocks this via lightweight refusal steering, which preserves the KL-optimal target while increasing the acceptance rate. Experiments on DeepSeek-R1-Distill and Qwen3 show ThinkSafe significantly improves safety while preserving reasoning proficiency, and achieves superior safety and comparable reasoning to GRPO with roughly an order of magnitude less compute. Code, models, and datasets are available at this [Github](https://anonymous.4open.science/r/ThinkSafe-E486/) and [HF repository](https://huggingface.co/neurips2026-10977).

Recent progress in latent world models (e.g., V-JEPA2) has shown promising capability in forecasting future world states from video observations. Nevertheless, dense prediction from a short observation window limits temporal context and can bias predictors toward local, low-level extrapolation, making it difficult to capture long-horizon semantics and reducing downstream utility. Vision-language models (VLMs), in contrast, provide strong semantic grounding and general knowledge by reasoning over uniformly sampled observation frames, but they are not ideal as standalone dense predictors due to compute-driven sparse sampling, a language-output bottleneck that compresses fine-grained interaction states into text-oriented representations, and a data-regime mismatch when adapting to small action-conditioned datasets. We propose a VLM-guided JEPA-style latent world modeling framework that combines dense-frame dynamics modeling with long-horizon semantic guidance via a dual-temporal pathway: a dense JEPA branch for fine-grained motion and interaction cues, and a uniformly sampled observation VLM thinker branch with a larger temporal stride for knowledge-rich guidance. To transfer the VLM's progressive reasoning signals effectively, we introduce a hierarchical pyramid representation extraction module that aggregates multi-layer VLM representations into guidance features compatible with latent prediction. Across EgoDex, EgoExo4D, BAIR Robot Pushing, and Physion, ThinkJEPA outperforms diverse latent world model and trajectory prediction baselines across egocentric trajectory prediction, long-horizon rollout, robotic latent prediction, and physical-scene forecasting. These results show that broad visual-semantic guidance from a VLM thinker can benefit JEPA-style latent forecasting.

Autonomous driving fundamentally demands dense understanding of visual, semantic, and geometric information, yet existing Vision-Language-Action (VLA) models are typically trained with sparse supervision such as language instructions or trajectory signals. This direct mapping from dense visual observations to sparse representations forces the model to compress or even discard critical scene information essential for safe driving. We argue that this limitation does not stem from insufficient model capacity, but from the lack of explicit mechanisms that encourage the formation of dense world representations in the latent space. In this spirit, we propose a novel paradigm for driving intelligence—Think Densely, Act Sparsely—and instantiate it with an efficient and interpretable framework, LECDrive (Latent Expert Cognitive Chains for Vision-Language-Action autonomous Driving). The core idea is to mimic the hierarchical perception process of humans by embedding a progressive chain of latent expert cognition within the VLA model. Concretely, we inject a compact set of task-specific latent tokens as carriers, enabling the model to internalize dense knowledge from foundational vision models, including visual representations (DINOv3), semantic structures (SAM3), and spatial geometry (DepthAnything3), while maintaining efficient inference. This interdependent cognitive chain follows a curriculum learning principle, where supervision is progressively structured from representation to semantics and geometry. Extensive experiments demonstrate that LECDrive consistently improves the performance of base VLMs across multiple autonomous driving benchmarks, covering key tasks such as object perception, state prediction, and trajectory planning. These findings suggest that constructing dense world models through structured latent expert chains is a promising direction for VLA-based autonomous driving.

Creating explorable 3D environments from text is important for immersive content creation, simulation, and interactive design. Recent methods often rely on image or video generative priors by synthesizing views along fixed or manually specified trajectories and lifting them into 3D. However, such passive pipelines cannot adapt view generation to the evolving scene memory, often leaving disoccluded regions incomplete and producing holes, floaters, and inconsistent geometry under free-viewpoint exploration. We argue that the key challenge is not to generate more views, but to decide where to generate useful views according to the current scene state. To this end, we propose ActPano3D, an active text-to-3D scene generation framework based on memory-guided panoramic exploration. ActPano3D formulates trajectory selection as spatial deliberation over the evolving scene memory, where candidate actions are evaluated by their expected utility for exploring unknown regions, repairing uncertain geometry, and improving the future 3D world model. The framework integrates expansion-refinement active planning, memory-routed panoramic generation, and reliability-aware scene memory fusion into a closed-loop generation-and-reconstruction process. The accumulated memory is further optimized into a renderable 3D Gaussian Splatting scene for free-viewpoint exploration. Experiments show that ActPano3D improves scene coverage, reduces rendering holes, and achieves better rendering quality and text alignment than existing text-to-3D scene generation baselines.

The task of visual reasoning relative pose identification (VRRPI) has exposed critical weaknesses in the multi-view 3D spatial reasoning capabilities of modern vision-language models (VLMs). We present a systematic analysis of VLMs behavior in VRRPI, uncovering stable output biases where VLMs disproportionately predict specific motion patterns regardless of visual evidence. To mitigate these inherent biases, we propose self knowledge re-expression with iterative debiasing (SKR-ID), an annotation-free pipeline that adapts VLMs for VRRPI using only randomly extracted unannotated image pairs. SKR-ID transitions VLMs' output mechanism from generic next-token prediction to task-optimized classification heads. The pipeline proceeds in two stages: (1) a cold-start initialization using debiasing object-centric prompts to anchor motion inference in 2D visual evidence; and (2) a recursive logit-ranking mechanism that enforces statistical parity by using the median logit value as a dynamic threshold. Experimental evaluations across frontier VLMs on standard benchmarks demonstrate that SKR-ID significantly mitigates systematic biases and unearths latent 3D reasoning ability. Our method achieves substantial gains, improving open-sourced VLMs' Macro-F1 scores by at least 8.3% and at most 23.0%, proving that VLMs possess significant untapped multi-view 3D reasoning potential.

Red teaming is essential for securing Large Language Models, yet current automated methods remain limited by templates and single-turn attacks. To simulate the complex, interactive nature of real-world adversarial attacks, we introduce a novel red teaming paradigm designed to maximize expected cumulative harm through strategic interaction. By formalizing red teaming as a Markov Decision Process in a hierarchical reinforcement learning framework, we navigate the challenges of sparse rewards and long-horizon planning. Our generative agent learns diverse, multi-turn attacks using a token-level harm reward, consistently uncovering vulnerabilities that bypass baselines. This approach achieves a new state of the art and reframes LLM red teaming as a principled, trajectory-based process.

Large language models (LLMs) are increasingly used for mental-health support; yet prevailing evaluation methods--fluency metrics, preference tests, and generic dialogue benchmarks--fail to capture the clinically critical dimensions of psychotherapy. We introduce THERAPYGYM, a framework that evaluates and improves therapy chatbots along two clinical pillars: fidelity and safety. Fidelity is measured using the Cognitive Therapy Rating Scale (CTRS), implemented as an automated pipeline that scores adherence to CBT techniques over multi-turn sessions. Safety is assessed using a multi-label annotation scheme, covering therapy-specific risks (e.g., failing to address harm or abuse). To mitigate bias and unreliability in LLM-based judges, we further release THERAPYJUDGEBENCH, a validation set of 116 dialogues with 1,270 expert ratings for auditing and calibration against licensed clinicians. THERAPYGYM also serves as a training harness: CTRS and safety-based rewards drive RL with configurable patient simulations spanning diverse symptom profiles. Models trained in THERAPYGYM improve on expert ratings, with average CTRS rising from 0.10 to 0.60 (and 0.16 to 0.59 under LLM judges). More broadly, THERAPYGYM provides foundational targets for responsible psychotherapy-oriented LLMs: clinical fidelity, therapy-specific safety, expert-calibrated evaluation, and multi-turn therapeutic process.