NeurIPS 2026 论文列表

Monocular depth foundation models generalize well across scenes, yet they are typically optimized with uniform pixel-wise objectives that do not distinguish user-specified or task-relevant target regions from the surrounding context. We therefore introduce Focusable Monocular Depth Estimation (FDE), a region-aware depth estimation task in which, given a specified target region, the model is required to prioritize foreground depth accuracy, preserve sharp boundary transitions, and maintain coherent global scene geometry. To prioritize task-critical region modeling, we propose FocusDepth, a prompt-conditioned monocular relative depth estimation framework that guides depth modeling to focus on target regions via box/text prompts. The core Multi-Scale Spatial-Aligned Fusion (MSSA) in FocusDepth spatially aligns multi-scale features from Segment Anything Model 3 to the Depth Anything family and injects them through scale-specific, gated conditional fusion. This enables dense prompt cue injection without disrupting geometric representations, thereby endowing the depth estimation model with focused perception capability. To study FDE, we establish FDE-Bench, a target-centric monocular relative depth benchmark built from image-target-depth triplets across five datasets, containing 252.9K/72.5K train/val triplets and 972 categories spanning real-world and embodied simulation environments. On FDE-Bench, FocusDepth consistently improves over globally fine-tuned DA2/DA3 baselines under both box and text prompts, with the largest gains appearing in target boundary and foreground regions while preserving global scene geometry. Ablations show that MSSA's spatial alignment is the key design factor, as disrupting prompt-geometry correspondence increases AbsRel by up to 13.8%.

A foundational challenge in neuroscience and AI is understanding how physical space is mapped into neural representations. While artificial neural networks can generate brain-like spatial representations, such as those observed in place and grid cells, their "black-box" nature makes it difficult to determine if these representations arise as general solutions or as artifacts of a chosen architecture, objective function, or training protocol. Critically, these models offer no guarantee that learned solutions for core navigational tasks, like path integration (updating position from self-motion), will generalize beyond their training data. To address these challenges, we introduce a first-principles framework based on the exponential map. Bypassing gradient-based optimization entirely, we use generator matrices to map physical locations to neural population vectors. Within the proposed formalism, essential navigational capabilities emerge analytically as intrinsic geometric properties. Exact, trajectory-invariant path integration naturally dictates commuting generators. Furthermore, skew-symmetric generators inherently yield equinorm representations with translationally invariant similarities. Translational similarity invariance is critical for stable egocentric navigation in open fields and empirical models typically enforce them via explicit regularization or architectural constraints. Extending our geometric perspective, we demonstrate that preserving the metric of flat space restricts the spatial wavevectors (derived from the generator eigenvalues) to form sets of roots of unity. The proposed framework supports diverse, biologically relevant spatial tuning profiles, including periodic grid fields, localized place fields, and context-dependent remapping. Beyond explaining mechanisms that underpin recent deep learning models of spatially-tuned cells, our work provides a formal connection between continuous attractor dynamics and oscillatory interference models. Finally, we show how the framework natively supports goal-oriented, multi-map navigation by identifying superpositions of representations alongside remapping as hyperdimensional computing operations. By grounding spatial representations in strict algebraic laws, this work offers a transparent alternative to empirical representation learning, revealing the exact conditions required for a coherent neural map of space.

Learned representations (or embeddings) are the foundation of modern machine learning systems, yet they are typically trained as fixed-size embeddings, without accounting for varying downstream resource or task constraints. Matryoshka Representation Learning (MRL) addresses this limitation by learning nested representations, but requires costly end-to-end retraining and does not support post hoc expansion, while Contrastive Sparse Representations (CSR) reduce training overhead via lightweight adaptors but operate at a 4x larger dimensionality. Additionally, all prior works restrict their framework towards compression only. In this work, we introduce \textbf{H}yperbolic \textbf{E}lastic \textbf{R}epresentation \textbf{L}earning (HERL), a framework for learning dimension-adaptive embeddings (compression and expansion) through a lightweight adaptor without retraining the base encoder. Our key idea is to exploit the exponential capacity of hyperbolic geometry by learning a non-linear mapping from the encoder space to hyperbolic space. We train a simple MLP to downsample the representations before projecting them onto a Lorentzian and learn adaptive representations that respect the Lorentzian geometry. Our approach is model-agnostic and effective in both supervised and unsupervised settings. Through extensive experiments across image and text modalities, we show that HERL preserves downstream performance across a wide range of embedding sizes, exhibits smooth interpolation between dimensions, and achieves these benefits at a fraction of the computational cost.

The Segment Anything Model 3 (SAM 3) has achieved significant progress in *Promptable Concept Segmentation (PCS)* by processing short noun phrases to generate segmentation masks with unique instance identifiers. While SAM3 excels at single-concept segmentation, it falls short in complex, real-world environments where multiple concepts must be segmented simultaneously. In particular, SAM3 typically fails to simultaneously process concepts with semantic overlap, leading to semantic misclassifications and redundant mask predictions. Moreover, as the number of concepts increases, the independent inference paradigm introduces prohibitive inference latency. To address these limitations, we propose the **D**ense **C**ompetitive **R**epresentations **SAM** (**DCR-SAM**) framework. By decoupling cross-modal interactions, the proposed method alleviates the inference latency. To resolve semantic misclassifications, DCR-SAM dynamically aligns textual concepts with visual representations and incorporates a dense competition head to enforce explicit inter-class competition. Furthermore, a learnable background token is applied to absorb non-target objects dynamically. These mechanisms effectively suppress overlapping predictions and generate high-fidelity masks. Extensive experimental evaluations across multiple benchmarks demonstrate that the proposed framework: I) achieves a 2.3\% $\sim$ 10.7\% performance improvement over state-of-the-art methods, II) delivers a 33$\times$ inference speedup compared to the native SAM 3, and III) exhibits strong robustness across diverse scenarios. All code will be released.

Electroencephalography (EEG) foundation models increasingly rely on multi-dataset training and evaluation, yet public EEG datasets still lack a shared task specification layer that can turn heterogeneous recordings into reusable benchmark units. Existing standards organize files, metadata, and provenance, but they do not specify EEG tasks under a common language and rulebook, leaving critical task semantics scattered across papers, code, and manual interpretation. We investigate whether heterogeneous public EEG datasets can be standardized through a structured task specification language paired with a shared rulebook. Our methodology represents each benchmark entry as a task document synchronized with an executable task kernel, with the rulebook defining task fields, evidence requirements, document-kernel alignment, review states, and machine-checkable constraints. Using this methodology, we release a community-reviewed EEG benchmark corpus centered on 53 completed and reviewed entries with 245 task definitions spanning diverse paradigms, and we introduce NeuroDoc and NeuroAudit as the operational support layer for rulebook-guided drafting, upgrading, review, amendment, and release management. We further examine whether the resulting benchmark units can be instantiated in a shared downstream setting across four EEG foundation model backbones, providing execution-based evidence for reusable, auditable, and executable EEG benchmarking infrastructure. Project resources are available at https://anonymous.4open.science/r/anonymous_reviewe_neurodoc.

Driver attention prediction is crucial for interpretable driver monitoring and human-like autonomous driving. Existing methods typically formulate this task as a scene-to-heatmap regression problem, failing to reveal the underlying dynamic attention redistribution mechanism. To address this issue, we propose a scene-task-aware competitive allocation framework, which decomposes driver attention as a bottom-up visual saliency prior and a top-down scene-task utility representation. An adaptive product-of-experts allocation rule is designed to fuse these two heterogeneous branches in a competitive manner, where scene-task demand dynamically modulates the dependence on top-down task utility, and driving control urgency adjusts spatial attention concentration via a learnable attention budget temperature. To explicitly differentiate task-critical regions from visually prominent yet task-irrelevant distractors, we incorporate competition-aware contrastive learning to enhance feature discrimination in the latent space. Extensive experiments demonstrate that our method achieves strong and balanced performance, with a Kullback–Leibler divergence of 0.98 and a normalized scanpath saliency of 4.15 on the DR(eye)VE dataset, as well as a Pearson’s correlation coefficient of 0.67 and a histogram similarity of 0.54 on the BDD-A dataset. Ablation studies and diagnostic analyses further validate that our framework yields interpretable attention allocation behaviors, including urgency-induced spatial entropy contraction and scene-task-demand-guided dynamic attention reallocation.

Multi-task reinforcement learning (MTRL) aims to train a single agent to efficiently optimize performance across multiple tasks simultaneously. However, jointly optimizing all tasks often yields imbalanced learning: agents quickly solve easy tasks but learn slowly on harder ones. While prior work primarily attributes this imbalance to conflicting task gradients and proposes gradient manipulation or specialized architectures to address it, we instead focus on a distinct and underexplored challenge: \emph{imbalanced data allocation}. Standard MTRL allocates an equal number of environment interactions to each task, which over-allocates data to easy tasks that require relatively few interactions to solve and under-allocates data to hard tasks that require substantially more experience to solve. To address this challenge, we introduce Distributionally Robust Adaptive Task Sampling (DRATS), an algorithm that adaptively prioritizes sampling tasks furthest from being solved. We derive DRATS by formalizing MTRL as a feasibility problem from which we derive a minimax objective for minimizing the worst-case return gap, the difference between a desired target return and the agent's return on a task. In benchmarks like MetaWorld-MT10 and MT50, DRATS improves data efficiency and increases worst-task performance compared to existing task sampling algorithms.

In-context learning (ICL) allows large language models (LLMs) to adapt to new tasks through demonstrations, yet it suffers from escalating inference costs as context length increases. While task vectors offer a promising alternative by compressing demonstrations into compact hidden-state representations, their quality has been evaluated only through downstream task accuracy. This indirect criterion provides limited insight into how to design more effective task vector extraction methods. In this paper, we posit that inference using task vectors should align their predictive distribution with that of ICL. To quantify this, we introduce $d_{\text{NTP}}$, a metric that measures the discrepancy in next-token probabilities between task vector-based and ICL-based inference. Our empirical analysis reveals that $d_{\text{NTP}}$ serves as a performance proxy, exhibiting a strong negative correlation with downstream accuracy. Motivated by this, we develop Linear Task Vector (LTV), a method designed to minimize $d_{\text{NTP}}$ via a closed-form linear mapping that estimates demonstration effects through regression. Across eight classification benchmarks and five LLMs, LTV consistently outperforms existing task vector baselines, improving average accuracy by 9.2\% while reducing inference latency. We further show that LTV outperforms the baselines on regression tasks. Moreover, we investigate the transferability of LTV across different model scales; an aspect that has remained nascent in task vector research. Specifically, we empirically show that task vectors from a larger model can enhance a smaller model's performance by 6.4\%, suggesting a new utility for extracted task representations.

Multi-Task Learning (MTL) is a fundamental problem in machine learning that has been extensively studied over the past decade. Recently, a variety of optimization-based MTL approaches have been proposed to jointly learn multiple tasks by modifying the optimization trajectory. In this paper, we argue that the design of mainstream optimization-based MTL methods implicitly assumes compatibility with momentum-free optimizers, which may limit the understanding of their effectiveness in modern training regimes. In practice, the instantaneously derived gradients from MTL operations contribute only marginally to the final parameter updates when momentum is present, resulting in an amortized rather than instant de-conflicting effect. Moreover, this amortized behavior can be further degraded under high-curvature optimization dynamics. To address this issue and enable the effective integration of mainstream MTL methods with advanced optimizers (e.g., Adam and Muon), we propose \texttt{APT} (Applicability of advanced oPTimizers), a lightweight framework featuring an adaptive momentum mechanism that balances the trade-off between instant and amortized de-conflicting. Furthermore, we show that the Muon optimizer can be interpreted as an implicit MTL learner, and we introduce a lightweight direction preservation strategy to better align with its orthogonalization process. Extensive experiments across four standard MTL benchmarks demonstrate that \texttt{APT} consistently enhances existing MTL approaches, yielding substantial performance gains.

Existing multi-task active learning (MTL-AL) methods predominantly rely on the linear scalarization of heterogeneous objectives, such as uncertainty, diversity, and gradient conflict.We argue that this entangled formulation is fundamentally limited, as linear combination fails to capture Pareto-optimal trade-offs under conflicting objectives. To address this limitation, we propose a paradigm shift from combination to deconstruction and introduce a principled framework that rethinks MTL-AL along three orthogonal dimensions. First, we identify the Conflict Paradox: contrary to conventional wisdom, samples with high gradient conflict are not noise to be avoided but carry maximal information about unresolved inter-task trade-offs. We therefore decouple sample selection, which embraces conflict, from optimization, which resolves it via gradient surgery. Second, we introduce Orthogonal Decomposition, separating the acquisition objective into two distinct axes: Informativeness and Exploration. This two-stage process, consisting of subspace projection followed by manifold coverage, prevents ``diversity dilution'' by excluding uninformative outliers before enforcing diversity. Third, we propose Cross-Task Mutual Information (CTMI), a distribution-free proxy for inter-task dependency derived from a variational lower bound under the Maximum Entropy principle. Extensive experiments on COCO, Cityscapes, and NYUv2 — including comparisons against recent multi-task active learning baselines — demonstrate consistent improvements over state-of-the-art MTL-AL methods, highlighting that structural rethinking of acquisition yields larger gains than incremental heuristic design. Our code is publicly available at: \href{https://anonymous.4open.science/r/multitask_active_learning-1545}{\texttt{https://anonymous/CTMI}}

Single-cell multi-omics technologies jointly measure multiple modalities from the same cell, such as gene expression, chromatin accessibility, and surface protein abundance, providing a richer view of cellular state than any single modality alone. A central challenge is how to learn a multimodal cell representation that preserves the different types of information contained in these co-profiled measurements. Existing integration methods often emphasize either cross-modal correspondence, which benefits retrieval and imputation, or fused discriminative structure, which benefits classification, but these objectives can favor different aspects of the data. As a result, a representation optimized for one setting may fail to preserve information needed for another. We introduce scDecomp, a decomposed representation learning framework for co-profiled single-cell multi-omics data. Motivated by the concepts of redundancy, uniqueness, and synergy, scDecomp separates multimodal information into three role-specific components: a redundancy component ($R$) for information shared across modalities, a uniqueness component ($U$) for modality-specific information, and a synergy component ($S$) for interaction-dependent multi-omic signals. This design allows the learned representation to retain shared cell-state structure while also preserving modality-specific and cross-modal regulatory information. On single-cell multi-omics benchmarks, the $R$ component achieves a 5.0\% reduction in FOSCTTM compared with the strongest baseline, while $U+S$ improves cell type classification accuracy by 1.38\%. Branch-level analyses further show that the decomposed components support complementary aspects of representation quality, suggesting that structured decomposition provides a practical strategy for learning more informative multimodal cell representations.

Foundational visual features such as DINO have played a critical role across modern computer vision, and have recently become key components in multi-view feed-forward geometry estimators. In this work, we demonstrate that by re-distilling these multi-view models-their internal knowledge of 3D geometry-into a single-view estimator, we can obtain enhanced 3D consistent foundational features. Our key idea is to construct a multi-view teacher by fusing pretrained 2D foundation features with multi-view geometric features, and refining the fused representation with a discriminative ranking objective. Through our discriminative distillation framework, we enforce the learned features to be both 3D consistent and locally distinctive, while keeping them aligned with the feature space of the original foundation model to preserve the semantic structure of the pretrained representation. Consistency and local discriminability are critical for 3D computer vision problems such as forming semantic and geometric correspondences across images. To demonstrate the effectiveness of our method, we perform comprehensive experiments spanning multiple angles: direct feature analysis, dense prediction transfer, and explicit 3D lifting and rendering. Across these evaluations, our method consistently produces stronger 3D-aware foundation features that improve multi-view consistency and local discriminability while preserving the semantic transferability of the original representation.

Training and evaluation in multi-channel imaging (MCI) remains challenging due to heterogeneous channel configurations arising from varying staining protocols, sensor types, and acquisition settings. This heterogeneity limits the applicability of fixed-channel encoders commonly used in general computer vision. Recent Multi-Channel Vision Transformers (MC-ViTs) address this by enabling flexible channel inputs, typically by jointly encoding patch tokens from all channels within a unified attention space. However, unrestricted token interactions across channels can lead to feature dilution, reducing the ability to preserve channel-specific semantics that are critical in MCI data. To address this, we propose Decoupled Vision Transformer (DC-ViT), which explicitly regulates information sharing using Decoupled Self-Attention (DSA), which decomposes token updates into two complementary pathways: spatial updates that model intra-channel structure, and channel-wise updates that adaptively integrate cross-channel information. This decoupling mitigates informational collapse while allowing selective inter-channel interaction. To further exploit these enhanced channel-specific representations, we introduce Decoupled Aggregation (DAG), which allows the model to learn task-specific channel importances. Extensive experiments across three MCI benchmarks demonstrate consistent improvements over existing MC-ViT approaches.

We cast image segmentation refinement as a \textit{discrete} diffusion process. To solve this task we introduce Cyclic Discrete Diffusion (CDD), a novel forward–reverse formulation with a cyclic label topology that maps a discrete label set to the uniform distribution and back. Instead of injecting Gaussian noise as is done for continuous diffusion, CDD evolves segmentation masks through label jumps: at each step, every pixel either remains in its current class or jumps to the next label with a fixed rate. This jump-or-stay mechanism defines a finite-state Markov process with bounded forward and reverse steps, featuring analytically consistent and tractable reverse dynamics. CDD operates natively in a multi-class label space, enabling joint refinement of all classes in a single pass without requiring class-wise decomposition, as commonly needed in prior binary refinement approaches. The proposed formulation admits a finite and controllable number of refinement steps, for which we provide a theoretical characterization based on spectral gap analysis. We evaluate CDD across diverse segmentation tasks, including multi-class semantic, instance-level, and binary segmentation refinement. Our method consistently improves both region accuracy and boundary quality across multiple backbones and datasets.

Multi-task learning (MTL) aims to jointly optimize multiple related tasks to obtain shared representations within a single model. Recent research has focused on developing MTL algorithms to alter optimization dynamics through task re-weighting or gradient manipulation. However, in this paper, we first empirically identify that current MTL approaches are highly sensitive to the model initialization, complicating empirical evaluation and limiting practical reliability, which has been overlooked but appears to be critical. To better understand this phenomenon, we provide a local theoretical analysis that motivates the connection between early-stage MTL optimization behavior and curvature around initialization. Motivated by this analysis, we propose a Curvature-guided Parameter Initialization (CPI) approach tailored for MTL. Specifically, CPI performs a short warm-up phase to probe local curvature around the initial parameters and heuristically biases the initialization toward empirically favorable curvature profiles for subsequent multi-task optimization. The proposed approach is optimizer-agnostic, requires no modification to existing MTL algorithms, and can be seamlessly integrated as a plug-in prior to standard training. Extensive experiments on standard MTL benchmarks show that CPI improves the aggregate multi-task trade-off while remaining compatible with mainstream MTL methods.

Merging separately trained LoRA adapters is a practical alternative to joint multi-task training, but it often hurts performance. Existing methods usually treat the LoRA update $\Delta W = BA$ as a single object and do not distinguish the two LoRA matrices. We show that the main source of LoRA merge interference comes from the output-side matrix $B$. Across tasks, $B$ repeatedly uses a small set of shared directions, while $A$ remains much more task-specific. As a result, the merged adapter overemphasizes these shared directions, and task-specific information is lost. We propose Pico (Pre-merge interference calibration in output-space), a data-free method that calibrates $B$ before merge by downscaling over-shared directions and then rescaling the merged update. Pico plugs directly into existing merging methods such as Task Arithmetic, TIES, and TSV-M. Across eight different benchmarks from math, coding, finance, and medical domains, Pico improves average accuracy by 3.4-8.3 points over the corresponding base method and achieves the best overall average performance. Pico also enables merged adapters to outperform the LoRA trained with all task data. These results show that LoRA merging works better when the two LoRA matrices are treated separately.

Language agents are moving from single-user assistants to shared collaborators in workspaces, forums, and group chats. A shared agent observes interleaved messages from multiple users, maintains persistent context, and executes tool actions, creating a boundary problem absent from standard single-user agents: the agent must decide not only whether an instruction is safe, but which user or task it is allowed to govern. We identify cross-user poisoning (CUP), where an adversary injects a message into shared context that is later applied while the agent serves a benign user, causing unauthorized actions or responses outside the instruction's intended scope. We validate CUP on two deployed multi-user agents, Continua and ElizaOS, and introduce MURMUR, a framework for evaluating shared-context agents under concurrent multi-user interactions. Across Slack, Workspace, and Airline domains, CUP achieves high attack success, persists across later interactions, and remains substantially more effective than matched prompt-injection attacks. We evaluate boundary-scoping defenses and find that context summarization alone is insufficient, while task clustering and provenance prompting substantially reduce non-adaptive propagation. These results show that robust multi-user agents require explicit user/task scoping rather than generic input filtering or context compression.

Large language models from different families use different hidden dimensions, tokenizers, and training procedures, making behavioral directions difficult to compare or transfer across models. We introduce an anchor-projection framework that maps hidden representations from each model into a shared anchor coordinate space (ACS). Behavioral directions extracted from source models are projected into ACS and averaged into a canonical direction. For a new model, the canonical direction is reconstructed into its native hidden space using only anchor activations, without fine-tuning or target-specific direction extraction. We evaluate five instruction-tuned model families and ten behavioral axes. We find that same-axis directions align tightly across the Llama-Qwen-Mistral-Phi cluster on the ACS. This shared structure transfers to downstream tasks: held-out targets achieve \(0.83\) ten-way detection accuracy and \(0.95\) mean binary AUROC, while canonical steering induces refusal-rate shifts of up to \($\Delta$ = +0.46\) under distribution shift. Sensitivity analyses show that two source models and small anchor pools already suffice to approximate transferable directions. Overall, ACS provides a novel perspective on cross-family interpretability, revealing that representation-level transfer remains robust across model families. The code for this paper is available at \url{https://anonymous.4open.science/r/ACS-annoymous/}.

The quadratic assignment problem (QAP) is widely considered one of the most difficult NP-hard combinatorial optimization problems, and one that remains challenging for neural constructive methods. We introduce CIMP, a Constructive policy trained by Imitation from Multi-expert teachers and built around an explicit Pairwise pre-conditioning block. CIMP reaches a 11.02% mean optimality gap on the 133 QAPLIB instances with published reference values, across 5 fixed same-config seeds, in a single constructive pass — without local search or per-instance refinement — with inference times between 0.1 and 3.1 seconds per instance. The model is trained on 400 synthetic instances with n ≤ 25. The most recent published neural reference reporting full QAPLIB results is SAWT [Tan and Mu, ICML 2024], a learn-to-improve (L2I) method that iteratively refines an assignment per instance after training on 5,120 synthetic instances, and reports a 26.8% mean gap. CIMP operates in a strictly more constrained learn-to-construct (L2C) regime and from a much smaller training budget, yet attains a substantially lower mean gap. The pre-block builds a state-dependent facility–location representation before scoring, a hybrid context-biased decoder mixes learned content similarity with QAP-specific algorithmic bias terms, and a sequence-distribution distillation objective is derived from a diversity-filtered pool of heuristic teachers. Family-wise and size-conditioned analyses, supported by a secondary normalized score, indicate that the gain is broadly consistent across QAPLIB families and across instance sizes beyond the training range.