NeurIPS 2026 论文列表

Reinforcement learning (RL) is well known to be data-intensive, and recent works have proposed leveraging data from similar systems to improve sample efficiency. In this work, we focus on the collaborative Linear Quadratic Regulator (LQR) setting and study the problem of learning under heterogeneous system dynamics. Due to the presence of a heterogeneity-induced additive bias, most existing collaborative methods are suitable only for low heterogeneity regimes and exhibit suboptimal performance under large heterogeneity. Based on the principle of Thompson sampling (TS), we propose an algorithm that, by leveraging data from other agents, yields a \emph{personalized} controller for every agent without incurring any additive heterogeneity-induced bias, i.e., with sublinear Bayes regret even under large heterogeneity. Under high heterogeneity regimes, we establish that our algorithm incurs $\tilde{\mathcal{O}}( T^{1-0.5\epsilon}\sqrt{\Delta})$ cumulative regret, where $T$ denotes the time horizon, $\epsilon\in[0,1]$ is a user-defined personalization parameter, and $\Delta$ denotes an upper bound on the maximum dissimilarity among the agents' dynamics. Our algorithm and analysis also generalize the special case of no heterogeneity. In particular, under no heterogeneity, we establish that our algorithm incurs $\mathcal{O}(\sqrt{T/M})$ cumulative regret. From a distributed implementation perspective, our method incurs only logarithmic communication overhead. Additionally, our algorithm outperforms methods that do not personalize data from other agents and, in certain regimes, also outperforms methods that do not utilize any data from other agents.

Large language models (LLMs) are increasingly adapted inside privacy- and regulation-constrained institutions where local post-training data cannot be centrally pooled. Federated learning therefore becomes a natural mechanism for collaborative LLM adaptation, but it also raises a safety question that is not captured by either classical federated learning or standard centralized alignment: clients may share a model while still requiring different safe behaviors. We formalize this setting as \emph{personalized federated safety}, where safety correctness is conditioned on the target client's local policy rather than treated as a single global refusal rule. To make this problem transferable rather than benchmark-specific, we introduce a compact, interpretable client policy space and a policy-conditioned benchmark construction framework, then instantiate it with five representative client archetypes spanning regulated healthcare, legal compliance, financial safety, youth safety, and enterprise general safety. The resulting benchmark contains 1{,}500 client-conditioned training examples and 750 held-out evaluation examples. We further propose \emph{Safety-Aware Weighted LoRA} (SAW-LoRA), a lightweight add-on for federated LoRA fine-tuning that lets each client selectively absorb incoming global task updates according to estimated task benefit and local-safety risk. Across two task datasets and two backbones, this add-on substantially lowers client-specific ASR for the strongest local safety mechanisms, reaching ASR near $1$--$2\%$ while preserving high benign acceptance. These findings position personalized federated safety as a concrete research problem with a practical update mechanism for federated LLM adaptation.

Existing large language model (LLM)-based memory systems apply universal, static policies that overlook a fundamental reality: the contexts that are worth storing in memory are different across users. This misalignment wastes limited memory budget on transient interactions while failing to preserve critical context for long-horizon tasks. To address this gap, we investigate an underexplored question: can LLM-based memory systems learn personalized memory policies? We introduce PerMemBench, the first benchmark for evaluating personalized memory systems, featuring multi-year, multi-domain interaction histories across diverse user personas. We further present the first empirical study of memory personalization and propose simple baseline methods. Our empirical study confirms that personalization yields substantial retention gains when the user profile is exactly inferred, yet reveals that accurate profile inference remains an open and critical challenge.

Long-term memory is increasingly important for personalized AI agents, yet existing benchmarks and methods remain largely text-centric. Even when images are included, the user-specific information needed for later questions is typically recoverable from text alone, and most memory systems reduce image turns to generic captions. Yet images often carry personal information that text rarely states---both \emph{explicit} evidence, such as recurring user-associated entities, and \emph{implicit} evidence, such as latent user facts inferred from visual or multimodal cues. We introduce a benchmark for \emph{personal visual memory} that targets both forms of evidence, and propose \textsc{VisualMem}, a hybrid visual--text architecture that augments a text-memory backend with a structured personal visual memory module. Rather than collapsing images into captions, \textsc{VisualMem} uses conversational context to resolve identity, ownership, and durable user facts. Experiments show that \textsc{VisualMem} substantially outperforms prior memory systems on our benchmark while remaining competitive on standard text-memory benchmarks, indicating that personal visual memory is a distinct and important component of long-term memory for personalized AI agents.

Fine-tuning large language models on narrow data with harmful content produces broadly misaligned behavior on unrelated prompts, a phenomenon known as *emergent misalignment*. We propose that emergent misalignment involves *persona-model collapse*: deterioration of the model's internal capacity to simulate, differentiate, and maintain coherent personas. We test this hypothesis behaviorally using two metrics: moral susceptibility and moral robustness; computed as the cross-persona and within-persona variability of models' Moral Foundations Questionnaire responses under persona role-play, respectively. We evaluate four frontier models (DeepSeek-V3.1, GPT-4.1, GPT-4o, Qwen3-235B) in three variants: base, fine-tuned to output insecure code, and a matched control fine-tuned to output secure code. Across the four models, insecure fine-tuning produces a $55$% average spike in moral susceptibility, pushing all four insecure variants beyond the band observed across 13 frontier models benchmarked in prior work, with GPT-4o reaching more than twice its upper end. It also causes a $65$% average drop in moral robustness (equivalently, a $304$% surge in within-persona variability), while the secure control preserves susceptibility near the base and induces only a partial robustness loss. Complementing these metric shifts, insecure variants' unconditioned responses converge toward saturation near the scale ceiling, departing markedly from both base models' differentiated responses and those elicited when they role-play toxic personas. Taken together, these metrics provide a sensitive diagnostic for emergent misalignment and serve as behavioral evidence that it involves persona-model collapse.

Large language models interact with users through a simulated "Assistant" persona. While the Assistant is typically trained to be helpful, harmless, and honest, it sometimes deviates from these ideals. In this paper, we identify directions in the model's activation space - persona vectors - underlying several traits, such as evil, sycophancy, and propensity to hallucinate. We confirm that these vectors can be used to monitor fluctuations in the Assistant's personality at deployment time. We then apply persona vectors to predict and control personality shifts that occur during training. We find that both intended and unintended personality changes after finetuning are strongly correlated with shifts along the relevant persona vectors. A central contribution of our work is preventative steering, a novel training-time intervention that actively suppresses the acquisition of unwanted traits. By steering activations during the fine-tuning process itself, this method “immunizes” the model against negative personality shifts.Moreover, persona vectors can be used to flag training data that will produce undesirable personality changes, both at the dataset level and the individual sample level. Our method for extracting persona vectors is automated and can be applied to any personality trait of interest, given only a natural-language description.

Simulating human behavior with Large Language Models (LLMs) offers a scalable laboratory for social science and stress-testing AI products. However, limitations remain in matching LLM outputs against the full breadth of human diversity. Current techniques for creating synthetic humans typically optimize for density matching, which often enforces behavioral uniformity and overlooks the rare, consequential outliers. We argue that robust simulation requires shifting the focus toward support coverage: ensuring synthetic populations span the entire landscape of possible human traits, opinions, and preferences. This paper introduces Persona Generators, functions that can produce diverse synthetic populations tailored to arbitrary contexts. We apply an iterative improvement loop based on AlphaEvolve, using LLMs as mutation operators to evolve our Persona Generator code rather than the personas themselves. The optimization process produces lightweight functions that can automatically expand small descriptions into populations of diverse synthetic personas that are maximizing coverage along relevant diversity axes. We demonstrate that evolved generators substantially outperform existing baselines across six diversity metrics on held-out contexts. Furthermore, by successfully mitigating standard LLM mode collapse, our generated populations actually capture real human trait distributions better than baselines specifically designed to match human statistics.

We study fixed-policy evaluation for finite Markov chains that may be reducible and periodic. Classical evaluation methods with gain and bias decomposition are not always diagnostic: the gain records only invariant Ces\`aro averages, while persistent phase-dependent behavior is absorbed into the bias together with genuinely transient effects. We identify the real peripheral invariant subspace $\mathcal{K}(P)$ of the transition matrix $P$ as the source of this ambiguity. Quotienting by $\mathcal{K}(P)$ is the minimal exact quotient that removes all non-decaying modes and makes the remaining dynamics strictly stable. After choosing a gauge projection $\Pi$ with kernel $\mathcal{K}(P)$, the reward admits a unique decomposition $r = g_\Pi^\star + (I-P)v_\Pi^\star$, where $g_\Pi^\star$ is a persistent regime profile and $v_\Pi^\star$ is a gauge-fixed transient component. An exact comparison with classical normalized gain and bias shows that the new pair reallocates the same information so that all persistent modes are represented in $g_\Pi^\star$ and $v_\Pi^\star$ is transient. This decomposition reconstructs finite-horizon returns, recovers statewise average reward, admits a transient-cost interpretation, and yields a stable estimator under a generative model.

Diffusion models are central to generative modeling and have been adapted to graphs by diffusing adjacency matrix representations. The challenge of having up to $n!$ such representations for graphs with $n$ nodes is only partially mitigated by using permutation-equivariant learning architectures. Despite their computational efficiency, existing graph diffusion models struggle to distinguish certain graph families and their spectra, unless graph data are augmented with ad hoc features. This shortcoming stems from enforcing the inductive bias within the learning architecture. In this work, we leverage random matrix theory to analytically extract the spectral properties of the diffusion process, allowing us to push most of the inductive bias from the architecture into the dynamics. Building on this, we introduce the Dyson Diffusion Model, which employs Dyson's Brownian motion to capture the spectral dynamics of an Ornstein-Uhlenbeck process on the adjacency matrix. Furthermore, conditioned on the spectral dynamics, we formulate a Lie group diffusion, appropriately modeling the remaining degrees of freedom. Strikingly, the resulting learning problem becomes permutation invariant at the Lie algebra level. We demonstrate that the Dyson Diffusion Model learns graph spectra accurately and outperforms existing graph diffusion models.

Tensor Singular Value Decomposition (t-SVD) has achieved strong empirical performance in multi-view clustering (MVC) and its incomplete setting (IMVC). However, the success of this framework heavily relies on the implicit dependency of its Fourier transform on the sample sequence, which raises serious concerns regarding sample-permutation sensitivity. To address this issue, we first formulate the permutation sensitivity in multi-view clustering by providing rigorous definitions of invariance at both the outcome level and the representation level. Secondly, by revealing the potential structural risks associated with the fixed Fourier basis, we propose two plug-and-play strategies: (1) an Order-Restoration strategy (K-OR), designed to reconstruct a locally smooth signal favored by the Fourier transform; and (2) an equivariant t-SVD framework (Cor-SVD), which fundamentally resolves the sensitivity of the Fourier transform by replacing fixed bases with data-dependent, equivariant transforms. Finally, we design a controlled permutation evaluation protocol to verify the robustness of tensorial clustering models. Extensive experiments demonstrate the significant performance variability of existing t-SVD-based methods under sample permutations, whereas our approaches substantially improve their permutation robustness while maintaining competitive clustering performance. These findings reveal critical vulnerabilities in current tensorial clustering benchmarks and underscore the necessity of the proposed method for achieving robust performance and reliable evaluation.

Large-scale visual generative models have achieved remarkable performance. However, their high computational and memory costs make deployment challenging in resource-constrained scenarios, such as interactive applications and personal single-GPU usage. Post-training quantization (PTQ) offers a practical solution by compressing pretrained models without expensive retraining. However, existing PTQ methods still suffer from severe quality degradation under extremely low-bit settings. In this paper, we identify channel ordering as an important but underexplored factor in per-group quantization. In this setting, each contiguous group shares one quantization scale. When channels with very different statistics are placed in the same group, the scale can be dominated by outliers and cause large quantization errors. Based on this observation, we propose PermuQuant, a simple and effective PTQ framework for low-bit diffusion models. PermuQuant sorts channels by a joint second-moment criterion before per-group quantization, placing channels with similar activation and weight statistics into the same group. It further uses a calibration-based acceptance rule to apply reordering only when the selected permutation reduces quantization error on calibration data. The selected permutations are absorbed into adjacent modules or applied to weights offline, avoiding explicit runtime permutation operations. Extensive experiments on multiple large diffusion models show that PermuQuant consistently reduces quantization error and outperforms existing PTQ baselines. On FLUX.1-dev with an NVIDIA RTX 5090, PermuQuant achieves up to a 1.8$\times$ single step speedup and reduces the DiT memory footprint by 3.5$\times$ under W4A4 NVFP4 quantization.

Large language models (LLMs) are increasingly deployed in enterprise settings where they handle sensitive documents and user context, raising acute concerns over security and controllability. Conventional access control regulates whether information is accessible to the model, yet leaves how the model uses that information at generation time largely unconstrained: once sensitive content enters the context, outputs may still drift beyond a user's authorized scope. We present Permit, a novel permission-aware representation intervention framework that closes this gap by enforcing fine-grained control directly on the model's hidden states. Through exploratory analysis, we find that permission conditions induce hidden-state shifts that are (i) separable across permissions and (ii) concentrated in a small set of dominant directions. Permit exploits this geometry in two stages: it first identifies a permission-sensitive subspace from activation differences across permission conditions, and then performs lightweight interventions within this subspace to steer generation, with two concrete instantiations (offset-based and gated). Both operate atop a frozen backbone with only a handful of permission-specific parameters, achieving precise control with minimal overhead. Experimental results demonstrate that \textsc{Permit} performs better than the state-of-the-art method across multiple permission settings while driving information leakage to near zero, achieving over \(18\%\) F1-score improvement with \(>98\%\) fewer trainable parameters.

Consistent video generation under editing operations requires persistence: when edits modify scene appearance or layout, subsequent generations should remain coherent across time and viewpoints. However, existing memory designs struggle to maintain long-term consistency after such modifications, as stored contexts may become outdated or invalid. To address this, we propose PermaVid, a novel framework built upon a multi-modal context memory that disentangles spatial context into semantic appearance and geometric structure, together with an edit-aware memory update and retrieval strategy that keeps memory evolution aligned with subsequent observations. Specifically, we develop two complementary memory banks: an RGB context memory that captures appearance-aware observations while implicitly encoding geometry, and a depth context memory that preserves geometry-only structure disentangled from semantics. Building on this design, we introduce a memory-guided video generation model that performs multi-modal feature fusion under reference conditions drawn from mixed-modality memory contexts. Experiments demonstrate that our method maintains strong long-term semantic and structural consistency after edits, significantly outperforming state-of-the-art methods.

Retrieving unknown atomic structures from observable analytical spectra or images remains a long-standing challenge across natural sciences. However, retrieval accuracy of existing retrieval methods on analytical data remains suboptimal because they have overlooked the underlying periodic quantum mechanical perturbations of non-equilibrium atomic structures behind analytical data. This paper proposes a periodic complex stochastic process (PCSP) that models such periodic perturbations and establishes theoretical backgrounds of periodic stochastic process in the complex-valued domain, including its sample diversity, process length, and periodicity. Finally, we develop a complex-valued cross-modal retrieval (CVCR) by integrating PCSP with cross-modal retrieval frameworks. CVCR outperformed existing cross-modal retrieval methods in cross-modal retrieval tasks of real-world analytical chemistry. Moreover, CVCR achieved state-of-the-art retrieval accuracy in zero-shot settings on 40 million of molecules.

Accelerating stochastic gradient methods with classical momentum schemes, such as Polyak's heavy ball, has proven highly successful in training large-scale machine learning models, particularly when combined with the hardware acceleration of large mini-batch computations. Yet, the effect of classical momentum on stochastic mini-batch optimization has been poorly understood theoretically, with prior works requiring strong noise assumptions and extremely large mini-batches. In this work, we develop a general theory of stochastic momentum acceleration for optimizing over quadratics in the interpolation regime, a popular abstraction for studying deep learning dynamics which also includes classical methods such as randomized Kaczmarz and coordinate descent. Our framework encompasses both heavy ball and Nesterov-style momentum, allows for arbitrary mini-batch sizes, and makes minimal assumptions on the stochastic noise. In particular, we show that acceleration from classical momentum is directly proportional to the gradient mini-batch size (up to a natural saturation point), thereby enabling perfect parallelization of mini-batch computations. Our theory also provides a simple choice for the momentum parameter, which is shown to be effective empirically.

Perception for action suggests that representations of the world should be shaped not by visual fidelity alone, but by their relevance for actions. At the same time, latent JEPA-style world models advocate learning compact predictive states from high-dimensional observations to facilitate the prediction of future states, but end-to-end training of these models is nontrivial because representations may collapse if our only goal is to construct a latent state that is easy to predict. We show that a suitable inverse dynamics regularization addresses both issues: it is an effective anti-collapse mechanism that induces action-aligned representations. By forcing latent states to preserve information about the action underlying a transition, it biases the model toward the controllable degrees of freedom of the environment while discarding uncontrollable distractors. This yields stable latent world models trained end-to-end from offline, reward-free trajectories, without frozen encoders, exponential moving averages, or complex latent regularizers. Empirically, the learned latent spaces are compact, interpretable, and enable competitive planning performance across simple 2D and 3D control tasks.

Although Large Multimodal Models (LMMs) have achieved strong performance on general video understanding, their susceptibility to textual prior shortcuts during causal discovery has been recognized as a critical deficit. The underlying mechanisms of this phenomenon remain incompletely understood, as existing benchmarks only measure response accuracy without revealing the sources and extent of the deficit. We introduce ProCauEval, a perturbation-based evaluation protocol that shifts from outcome assessment to mechanism diagnosis, probing causal discovery through five controlled configurations that systematically manipulate visual and textual modalities to decompose their respective contributions to model behavior and dissect the failure modes. Evaluating 17 mainstream LMMs, we find that models faithfully perceive video content yet systematically underexploit it during causal reasoning. We further observe that stronger post-training amplifies rather than mitigates textual prior reliance, and that higher baseline performance correlates with greater fragility under perturbation. To address these, we propose Anti-Distillation Policy Optimization (ADPO), a reinforcement learning framework built on negative teacher alignment, which augments GRPO by explicitly pushing the policy away from a prior-only counterfactual teacher induced by visual corruption. Specifically, ADPO maximizes the divergence between the policy distributions conditioned on the original and visually corrupted inputs, thereby forcing the model to ground its reasoning in visual evidence rather than textual shortcuts. Extensive experiments show that ADPO improves visual engagement without sacrificing fundamental comprehension, thus offering a preliminary step toward reliable causal discovery.

Building structured 3D scene layouts from a single image requires reconciling visual observations with physical and spatial constraints, a challenge that is difficult to address with direct prediction alone. In this work, we formulate monocular 3D layout estimation as a perceive-then-plan problem with vision-language models, where a Perceiver first grounds the 3D objects and then a Planner iteratively refines the scene hypothesis through actions that improve physical plausibility while preserving consistency with the input image. We propose Layout-as-Policy (LaP), which casts the planning stage as a policy learning problem: 3D layouts are represented as structured states, and refined via discrete actions such as translation, rotation, and rescaling. Starting from an observation-aligned initialization with the geometry-enhanced Perceiver, the LaP Planner is trained to produce action sequences that progressively resolve geometric inconsistencies and enforce realistic spatial relations. To enable effective learning, we combine supervised trajectory initialization with preference-based optimization, allowing the model to learn corrective behaviors without requiring explicit reward engineering. This formulation transforms layout estimation from a one-shot prediction task into an iterative refinement process, enabling better handling of global constraints and complex object interactions. Experiments demonstrate that our approach produces layouts that are more physically coherent and better aligned with visual observations, while naturally supporting downstream tasks such as scene editing and manipulation.

While recent vision-language models (VLMs) demonstrate strong multimodal understanding, they remain limited in spatial reasoning tasks that require active evidence acquisition and multi-step visual interaction. This limitation suggests that relying solely on implicit visual representations from vision encoders is insufficient for recovering fine-grained spatial evidence. We introduce *PERception-Interaction-reason Agent* (**PERIA**), a tool-augmented visual agent for spatial reasoning tasks across map reasoning, visual probing, and vision reconstruction. **PERIA** uses two lightweight tool families: *vision perception tools* for exposing textual, symbolic, and spatial evidence, and *vision interaction tools* for manipulating visual context, tracing paths, and verifying spatial relations. To train **PERIA**, we develop a unified recipe that combines supervised tool-use trajectory synthesis, composite rewards, and Observation-Relaxed Group-in-Group Policy Optimization (OR-GIGPO) for effective multi-tool behavior. Experiments on 13 benchmarks from 8 datasets show that **PERIA**-8B improves over the Qwen3-8B backbone by 10.0% on in-distribution benchmarks and 4.4% on out-of-distribution benchmarks, while outperforming previous state-of-the-art baselines of similar size by 7.0%–14.8%. It also achieves performance comparable to much larger models such as Qwen3-VL-235B-A22B-Thinking and GPT-5, demonstrating the effectiveness of **PERIA** in enhancing spatial reasoning capabilities.