NeurIPS 2026 论文列表

Integrating human feedback into language models has gained increasing attention as a means to align model outputs with human preferences, with Reinforcement Learning from Human Feedback (RLHF) serving as a prominent framework for preference-based alignment. Recent studies have incorporated eye-tracking (ET) data as an additional supervisory signal for reward model training, grounding reward learning on human reading behavior. Building on this approach, we extend prior fixation-centric models by leveraging established findings from cognitive psychology on human letter recognition. Specifically, we model visual attention as a spatially graded distribution centered on the current gaze location, and incorporate this representation into reward model training. Our results show that the reward model trained with the proposed gaze distribution achieves higher preference prediction accuracy over a baseline model. Moreover, the fitted attentional distribution reflects key properties of human attention during reading, suggesting that it captures cognitively meaningful aspects of attention allocation.

Latent visual reasoning involves visual evidence more directly in multimodal reasoning by inserting continuous latent tokens before textual generation. However, the necessity of these latent tokens at inference remains ambiguous. We show that replacing latent tokens with random noise or removing them completely causes little performance degradation across spatial reasoning benchmarks. Reinforcement learning further diminishes the latent generation behavior after post-training. These observations raise a central question: $\textit{Is latent visual reasoning still meaningful}$? We argue that its value should be measured by how effectively latent tokens guide learning, rather than whether they persist as an inference-time format. Our analysis shows that latent reasoning is unevenly favorable across question types, yet hard task-level routing for applying latent generation is brittle. Motivated by these findings, we propose an attention-based reward that encourages generated latent tokens to interact with later text tokens during RL. This reward promotes latent utilization when the latent mode is activated while preserving the flexibility to use pure-text reasoning. Experiments show that our method improves performance across perception and visual reasoning benchmarks, even when latent tokens are rarely generated after post-training. Our results highlight that, without explicit expression at inference, latent visual reasoning can shape better visual grounding and more accurate textual reasoning $\textbf{in silence}$. Our code and trained models are publicly available at Hugging Face https://huggingface.co/collections/doubleblindsubmission/submission3847.

The performance of machine learning models often relies on large labeled datasets; however, data collected from diverse sources can contain label noise. Recent work has shown that, in noisy settings, there may exist a subset of the training data on which models can achieve performance comparable to training on a noise-free dataset. A widely used method for identifying such subsets is \textit{cutstats}, which employs $k$-nearest neighbors ($k$-NN) to detect low-noise samples. However, its performance on high-dimensional data remains largely unexplored. In this work, we formally establish that the performance of a classifier trained on a subset of a noisy dataset selected via \textit{cutstats} is influenced by the accuracy of $k$-NN. We further demonstrate that, in noisy environments, exploiting data invariance and knowledge of underlying symmetries can significantly enhance the performance of $k$-NN, bringing it closer to the Bayes optimal classifier even in high-dimensional regimes. Finally, we show that for real-world scenarios, where information about the underlying invariance is only partially known, learnt invariant representations can still facilitate the identification of near-optimal subsets.

One of the fundamental differences between biological neural networks and artificial neural networks (ANNs) is that biological neurons obey Dale’s principle: each neuron projects either exclusively excitatory or exclusively inhibitory outputs. While this constraint is central to biological circuits, previous attempts to impose Dale’s principle in ANNs have often led to training instability or degraded performance. In this work, we introduce EISep RNN, a recurrent neural network architecture that implements Dale’s principle through a learnable excitatory–inhibitory separation. Across a series of learning settings, including multi-task learning, continual learning, and reinforcement learning, EISep achieves more stable training dynamics, competitive or improved performance, and sparser, more modular connectivity than vanilla RNNs and previous Dale-constrained models. EISep also exhibits stable and informative latent dynamics. These results establish Dale’s principle as a biologically grounded and computationally effective inductive bias for recurrent neural networks.

Humanoid robots capable of versatile whole-body interaction with everyday objects represent a central goal of embodied intelligence. Existing approaches often rely on motion-reference inputs or task-specific rewards, coupling policies to particular motion scripts, object geometries, and contact timings. This limits geometric generalization and flexible interaction control, where simple commands specify motion intent while object geometry determines how contact should be executed. We introduce LessMimic, a framework for versatile motion-free-at-inference humanoid–object interaction. At inference, a single whole-body policy is steered by root commands and an interaction-type flag instead of motion-reference inputs, and is conditioned on a compact interaction representation built from short histories of distance-field-derived surface distances and local surface directions. This geometry-conditioned interface enables command-steered control while adapting contact behavior to local object geometry. Through visual distillation, LessMimic further enables egocentric-depth-based deployment for MoCap-free sensing. Across object scales from 0.4× to 1.6×, LessMimic maintains robust performance over four interaction tasks, including PickUp, SitStand, Push, and Carry, where motion-conditioned baselines degrade sharply away from the training scale. Beyond single-task evaluation, LessMimic further supports sequential composition of heterogeneous interaction skills, attaining 62.1% success on five-task sequences and retaining 23.5% success at 15 task instances. By grounding interaction control in local geometry rather than motion references, LessMimic provides a path toward geometry-generalizable, command-steerable humanoid–object control with heterogeneous skill composition.

AI-Generated Image (AIGI) Detection is becoming critical as generative models proliferate, yet existing detectors often generalize poorly by relying on narrow, dataset-specific artifacts. Many prior works improve generalization through additions, such as introducing extra artifact cues, pretraining models, training constraints, or test-time operations, which may increase inference cost and model complexity. In this paper, we take the opposite route and study generalization through subtractions and propose AID--, a plug-and-play framework that improves transformer-based AIGI detectors from both feature and parameter perspectives. At the feature level, we introduce high-order dependency ranking to identify and prune redundant or potentially misleading tokens for progressive token reduction; and at the parameter level, we design relative-advantage head supervision and a class-asymmetric exit rule to train depth-specific early-exit heads. AID-- can be seamlessly integrated into transformer-based AIGI detectors to improve both efficiency and out-of-distribution generalizability, which is experimentally validated on six public datasets. For example, AID-- helps the detector Effort achieve up to a 2.2× speedup while improving generalization accuracy by 3.6% on GenImages.

Localizing AI-edited regions is essential for interpretable forensic analysis, but remains challenging due to subtle and spatially distributed artifacts that are misaligned with semantic or object boundaries. Existing approaches rely on pixel-level supervision from controlled editing pipelines, which is difficult to scale and can introduce misleading signals: artifacts frequently extend beyond annotated regions, while out-of-mask pixels are treated as authentic. This limits models' ability to capture transferable evidence and generalize across generators and datasets. To address these issues, we propose \textbf{ReGFLoW}, a \textbf{Re}construction-\textbf{G}uided \textbf{F}ake \textbf{Lo}calization framework under \textbf{W}eak supervision, which is the first weakly supervised approach for diffusion-edited fake region localization. ReGFLoW requires only real/fake labels at the image-level and uses diffusion reconstruction errors as dense spatial guidance to inject them into both feature and score spaces. Furthermore, by artifact-centric multiple instance learning, ReGFLoW utilizes localized diffusion evidence without relying on semantic-affinity or boundary-based pseudo-mask priors. Extensive experiments demonstrate that ReGFLoW achieves stronger out-of-domain generalization than fully supervised learning baselines.

Many real-world applications require highly interpretable machine learning systems, particularly in high-stakes domains such as healthcare. A promising recent direction is the paradigm of Maximal Coding Rate Reduction (MCR$^2$), which seeks to characterize and preserve the low-dimensional structure underlying each class. However, we find that exhaustively preserving structural information is often unnecessary and can even be detrimental in supervised learning, especially under noisy and uncertain real-world conditions, as it may overly constrain the fitting flexibility of the model. In contrast, we propose a simple yet effective framework, termed SimCoding, which captures only minimal structural information for each class. Despite relying on substantially less structural information, SimCoding significantly improves the generalization ability and robustness of deep models. Extensive experiments across diverse benchmark datasets demonstrate the effectiveness of SimCoding. We further validate SimCoding on a challenging real-world application, Parkinson’s disease severity assessment from free-living human activity signals, where it consistently achieves strong robustness and outperforms competing methods. Moreover, SimCoding can be naturally extended to incremental learning scenarios, where it substantially alleviates catastrophic forgetting and consistently surpasses strong baselines.

Vision-language-action models (VLA) promise human-steerable autonomous driving, but their training is bottlenecked by the scarcity of frames paired with natural-language instructions: while camera streams and expert trajectories are logged at scale, language annotations (e.g., "turn left at the intersection") remain scarce and expensive to acquire. To address this challenge, we introduce Latent Action Driving Annotations (LADA), a three-stage pipeline that transforms abundant unlabelled observation–trajectory pairs into a substrate for language-conditioned control. First, we train a latent action model with a vector-quantised bottleneck, producing a compact codebook of high-level vehicle intents. Second, a small language-annotated subset is used to train a vision-language translator to map observations and language instructions into this codebook. Third, we train a driving VLA on observation–latent-action pairs over the full unlabelled corpus. Using fewer than 5% of language annotations and without leveraging any auxiliary chain-of-thought reasoning or visual question answering streams, LADA achieves a Driving Score of 87.98 and a Success Rate of 70.46% on the closed-loop Bench2Drive benchmark, matching or surpassing fully supervised baselines.

Evidence-grounded medical question answering requires not only accurate answers, but also precise, verifiable, and clinically trustworthy evidence. Existing medical RAG systems typically retrieve top-\(k\) chunks according to item-wise relevance. Although this design improves recall, it does not explicitly optimize whether the selected evidence set is compact, sufficient, and decision-critical. In this paper, we study medical evidence-grounded QA from a different perspective: instead of retrieving evidence without optimization, the evidence-grounded medical QA system should identify a compact sufficient subset of high-density evidence snippets. We propose **MEGA**, a gain-aware framework for compact supporting medical evidence selection. MEGA first expands clinical queries into complementary subqueries and performs hybrid retrieval to construct a recall-oriented candidate evidence pool. It then introduces hidden-state **Information Gain Scoring**, which uses a frozen LLM to estimate whether each candidate snippet contributes new answer-relevant information beyond surface relevance. Finally, MEGA formulates evidence selection as a budgeted utility maximization problem and proposes an **FPTAS selector** to identify a compact evidence subset under an explicit token budget. We further construct two evidence-grounded medical QA benchmarks, CRC-EvidenceQA and Med-EvidenceQA, to evaluate both answer quality and evidence grounding. Across three LLM backbones and five medical RAG baselines, MEGA consistently achieves the best results, improving over the second-best baseline by \(8.8\%\) on average in answer quality and 38.8\% on average in evidence F1. Ablation studies and blinded clinical expert evaluation further validate its robustness and clinical relevance.

This paper examines the role of Novel View Synthesis (NVS) in geometric representation learning. In principle, NVS should reason about 3D scene structure, thereby enabling transferable multi-view geometric representations. Yet, existing encoder-based NVS methods yield poor representations. This is not because of lack of supervisory signal, but rather due to inconspicuous architectural choices: spatially expressive decoders that dilute representational capabilities of the scene encoder, and low-level pixel-space targets that hinder feature learning. We present SNAP, a self-supervised encoder-decoder transformer that addresses both through a pose-conditioned local decoder and a latent-space reconstruction objective. SNAP is general purpose, and we show that it is on par or better than special-purpose geometry-supervised methods. SNAP also outperforms self-supervised representations across five geometric tasks: visual localization, pose estimation, point correspondence, depth estimation, and robot manipulation. Remarkably, SNAP's patch tokens exhibit emergent viewpoint invariance that rivals heavily supervised models despite lower compute and data budgets. Under severe camera shifts where standard 2D representations collapse, SNAP maintains strict spatial consistency, revealing that restricting decoder expressivity actively prevents the suppression of transferable geometric structure.

Vision Language Models (VLMs) offer the exciting possibility of processing text as rendered images, bypassing the need for tokenizing the text into long token sequences. Since VLM image encoders map fixed-size images to a fixed number of visual tokens, varying rendering resolution provides a fine-grained compression knob. However, accuracy deteriorates quickly as compression increases: characters shrink below the vision encoder's effective resolution, making them indistinguishable. To address this, we propose LensVLM, an inference framework and post-training recipe that enables VLMs to scan compressed images, then selectively expand only the relevant images to their uncompressed form via learned tools. Building on Qwen3.5-9B-Base, LensVLM maintains accuracy comparable to the full-text upper bound at 4.3$\times$ effective compression and outperforms retrieval-based, text- and visual-compression baselines up to 10.1$\times$ effective compression across seven text QA benchmarks. LensVLM also generalizes to multimodal document and code understanding tasks, with the accuracy gain over baselines growing as compression increases. Our analysis validates this approach: training makes visual compression robust to rendering choices, and as compression grows the model increasingly relies on expanded content rather than unreliable visual reading. The analysis also yields practical tool-choice guidance: text expansion is preferable for rendered text, while high-resolution image expansion suits native documents whose layout cues carry task-relevant information.

LLMs are increasingly used not only to generate text from scratch, but also to polish and extend human writing. This makes AI detection a fine-grained problem: downstream users often need to understand the level and form of AI involvement in a text, rather than merely separate human from machine. Existing detectors are poorly suited to this setting because they compress the evidence into scalar scores, selected features, or surface linguistic cues, which miss the subtle patterns left by different forms of AI involvement. We propose ***LensCT***, which makes a frozen LLM's internal predictive behavior visible as a tomogram: an image-like map of how the model responds across tokens and layers. The tomogram captures both temporal patterns along the input sequence and hierarchical patterns across transformer layers, providing a richer view of AI involvement than compressed detector scores. A vision backbone then reads this tomogram to classify the level of AI involvement. On a **237,500**-sample English benchmark spanning five domains and six frontier LLM generators, ***LensCT*** achieves **0.990** in-distribution macro-AUROC, reducing error by **28.6\%** over the strongest baseline. Across **697,500** total evaluation samples, ***LensCT*** consistently outperforms prior detectors under cross-generator, cross-domain, cross-language, and adversarial shifts, showing that temporal-hierarchical tomograms provide a robust and transferable signal for fine-grained AI-involvement detection.

Large language model agents now combine tools, long context, memory, and multi-step workflows, yet they still fail when execution loses track of the evidence required for a reliable answer. Common failure modes include missing support, unresolved contradictions, and evidence lost under context pressure. We present Lemon, an evidence-risk-aware framework that makes such risks an explicit control variable. At the intra-instance level, a single state-conditioned controller jointly schedules reasoning, tool invocation, worker expansion, verification, context compression, and memory operations. A recoverable evidence substrate pairs anchored context compression with reusable semantic memory, preserving raw tool outputs while storing transferable process fragments. We further extend Lemon to an inter-instance setting in which heterogeneous personalized agents expose expertise, exchange evidence-bearing proposals, critique unsupported claims, and internalize useful collaboration artifacts. Empirically, Lemon reaches 91.36\% accuracy on GAIA and 80\% on xbench-DeepSearch. In an open-source reproducibility comparison on GAIA, Lemon uses 49.9\% to 88.4\% fewer tokens per task than three top-ranked agent baselines.

Scalar quantization of large language models (LLMs) is fundamentally limited by information-theoretic bounds. While vector quantization (VQ) overcomes these limits by encoding blocks of parameters jointly, practical implementations must avoid the need for expensive lookup mechanisms or other explicit codebook storage. Lattice approaches address this through highly structured and dense packing. This paper explores the Leech lattice, which, with its optimal sphere packing and kissing configurations at 24 dimensions, is the highest dimensional lattice known with such optimal properties. To make the Leech lattice usable for LLM quantization, we extend an existing search algorithm based on the extended Golay code construction, to i) support indexing, enabling conversion to and from bitstrings without materializing the codebook, ii) allow angular search over union of Leech lattice shells, iii) propose fully-parallelisable dequantization kernel. Together this yields a practical algorithm, namely Leech Lattice Vector Quantization (LLVQ). LLVQ delivers state-of-the-art LLM quantization performance, outperforming recent methods such as Quip\#, QTIP, and PVQ. These results highlight the importance of high-dimensional lattices for scalable, theoretically grounded model compression.

Web agents promise to automate consequential operations on enterprise web systems, yet the benchmarks used to evaluate them score correctness on surface-level proxies: DOM/URL predicates, action-trace matchers, or vision-language verdicts on a closing screen. These proxies are inadequate for enterprise resource planning (ERP) systems, where every action commits to a transactional ledger and the criterion of correctness is what the ledger now records, not what the page now renders. We argue that ERP web operations should be evaluated against what a run actually wrote to the relational backend, with the server-side operation trace recorded alongside the outcome to provide the audit provenance that enterprise governance requires. We introduce Ledger, a path-validated, database-grounded benchmark for browser agents on a production-grade open-source ERP: 424 oracle-certified task instances across 53 functional modules, scored by an Atomic Edit Set validator (field-level $F_1$ against a PostgreSQL oracle, under a typed normalization function) paired with a Canonical Operation Tracer (server-side instrumentation of the ERP’s JSON-RPC dispatch). Each episode runs under pg_restore snapshot isolation; no language-model judge is invoked; three formal guarantees (score determinism, episode isolation, tracer completeness) make scores auditable and exactly replayable. A cross-paradigm evaluation of seven contemporary browser agents shows Ledger discriminates architectures sharply, with task-success rate spanning 81.6% down to 13.7% across the roster while long-horizon enterprise workflows remain far from saturated.

Autoregressive video generation has achieved impressive results on short clips, yet generating long videos remains challenging due to error accumulation over extended horizons, where each predicted frame depends on potentially imperfect previous frames. Existing approaches typically rely on static context management strategies, such as sliding-window caches that discard early frames or fixed sink frames that permanently preserve initial content. These designs either lose critical long-range context or anchor generation to outdated information, which can lead to degraded fidelity, motion stagnation, and reduced diversity. In this paper, we propose \textbf{Learning to Memorize (L2M)}, a learned context memory management framework that dynamically retains, evicts, and evolves historical frames according to predicted importance. The framework consists of an \emph{importance prediction router} that estimates the relevance of each historical frame, \emph{degradation-aware training} that improves robustness to noisy or corrupted context, and a \emph{dynamic memory initialization and evolution} mechanism that updates long-term memory using exponential moving average (EMA) importance scores. This allows high-quality frames to replace stale anchors and enables memory to adapt naturally to scene evolution. Trained with a teacher-student distillation objective and sparsity regularization, L2M makes efficient use of a fixed KV-cache budget while preserving the most informative contextual information. Extensive experiments show that L2M achieves superior long-term video generation performance over existing baselines, improving stability and visual fidelity, and establishing a new paradigm for learned memory management in long autoregressive video generation.

Learning-augmented algorithms have emerged as a powerful paradigm to surpass traditional worst-case lower bounds by integrating potentially noisy predictions. While this framework has seen success in online scheduling, existing work primarily optimizes job latency while relying on frequent, "blind" preemptions. This ignores the fundamental trade-off between algorithmic performance and preemption complexity. We provide the first systematic study of learning-augmented scheduling that curbs preemption while optimizing latency. We establish that the gap between theoretical latency bounds and preemption overhead can be bridged with solid analytical foundations. Our results include $O(1)$-competitive algorithms for single and unrelated parallel machines with only $O(1)$ preemptions per job under accurate predictions, with overhead scaling logarithmically with the prediction error. By providing the first bounded-preemption guarantees for unrelated and malleable machines, we extend the theoretical reach of the learning-augmented framework to more constrained and realistic settings. Finally, our algorithms are validated through experiments.

We study learning-augmented online portfolio selection (OPS), where an investor uses predictions to improve wealth while remaining protected against unreliable advice. In frictionless markets, we characterize the optimal tradeoff between robustness (wealth under arbitrary predictions) and consistency (wealth under perfect predictions) tied to the geometric mean, and show that the optimal tradeoff admits a tractable online characterization. We further quantify how performance interpolates between these endpoints under imperfect predictions by smoothness guarantees. Under transaction costs, we give exact robust-trading certificates and, to our knowledge, the first learning-augmented OPS impossibility under costs: for any positive cost, no GM-robust policy can be maximally consistent under exact predictions. Lastly, we show that a cost-aware greedy rule remains Pareto-optimal.