论文检索

Optimization · Non-Convex

Zihao Zheng, Irwin King, Songtao Lu

Reinforcement learning (RL) is often hierarchical: an upper-level (UL) learner selects model parameters while a lower-level (LL) decision-making process responds, forming a nested two-level optimization structure captured by bilevel optimization. Most existing bilevel RL methods assume a single-policy LL Markov decision process (MDP), and thus miss the competitive structure in applications such as incentive design where multiple policies interact. We study a class of bilevel optimization problems whose LL is a regularized min--max zero-sum Markov game, and whose UL optimizes through the saddle-point equilibrium induced by the LL game. We propose a penalty-based first-order policy-gradient method built on the Nikaido–Isoda function, termed penalty-augmented Nikaido–Isoda descent–ascent (PANDA), which avoids UL hypergradients and requires no second-order information by exploiting the min--max game structure. We prove that PANDA converges to stationary points of this bilevel problem without restrictive assumptions such as convexity on either the UL or LL objectives. Moreover, PANDA reaches an $\epsilon$-stationary point in $\tilde{\mathcal{O}}(\epsilon^{-1})$ iterations with sample complexity $\tilde{\mathcal{O}}(\epsilon^{-3})$, matching the best-known rates for bilevel RL with single-policy LL MDPs. Experiments further demonstrate superior performance over closely related baselines.

Social Aspects · Everything Else

Xinyang Lu, Jiabao Pan, Rachael Hwee Ling Sim, See-Kiong Ng, Anthony Tung, Bryan Kian Hsiang Low

The rapid development of large language models (LLMs) has raised concerns regarding the inclusion of private or inappropriate data during training, which has led to growing interest in LLM unlearning. Many existing LLM unlearning approaches rely on prediction loss-based optimizations, such as maximizing the loss on the forget set. However, these methods often face issues such as over-forgetting and poor model utility. In this work, we address these issues by introducing a novel perspective that shifts the unlearning optimization target to reducing data attribution instead. We propose the first LLM unlearning framework based on data attribution rewards called DareU that employs reinforcement learning to update the LLM and reduce the attribution score of generated responses (i.e., de-attribute) to the forget data owners. Experimental results using an LLM classifier as an efficient approximation of attribution demonstrate that DareU outperforms existing baseline approaches, achieving effective unlearning while balancing forget quality and model utility.

zhaoyang zhang, Shuli Jiang, Yantao Shen, Yuting Zhang, Dhananjay Ram, Shuo Yang, Zhuowen Tu, Wei Xia, Stefano Soatto

Reinforcement learning (RL) post-training has recently driven major gains in long chain-of-thought reasoning large language models (LLMs), but the high inference cost of such models motivates distillation into smaller students. Most existing knowledge distillation (KD) methods are designed for supervised fine-tuning (SFT), relying on fixed teacher traces or teacher-student Kullback–Leibler (KL) divergence based regularization. When combined with RL, these approaches often suffer from distribution mismatch and objective interference: teacher supervision may not align with the student’s evolving rollout distribution, and the KL regularizer can compete with reward maximization and require careful loss balancing. To address these issues, we propose \emph{RL-aware distillation} (RLAD), which performs selective imitation during RL---guiding the student toward the teacher only when it improves the current policy update. Our core component,Trust Region Ratio Distillation (TRRD), replaces the teacher-student KL regularizer with a PPO/GRPO-style likelihood-ratio objective anchored to a teacher--old-policy mixture, yielding advantage-aware, trust-region-bounded distillation on student rollouts and naturally balancing exploration, exploitation, and imitation. Across diverse logic reasoning and math benchmarks, RLAD}consistently outperforms offline distillation, standard GRPO, and KL-based on-policy teacher–student knowledge distillation.

Applications · Computer Vision

Tingran Wang, Changshuo Wang, Pinjie Xu, ZhangHuang, Yuan Shi, Linjun Sun, Weijun Li

Three-dimensional (3D) reasoning segmentation aims to segment target objects based on text instructions and 3D spatial cues. Recent efforts in 3D reasoning leverage Multimodal Large Language Models (MLLMs) to bridge the gap between text and 3D data. However, since MLLMs are primarily trained on text-image pairs, directly adapting them to unstructured 3D point clouds often fails to capture implicit semantic intent and reliably localize objects. This paper introduces TVDRNet to address these challenges. Inspired by Active Vision theory, where humans selectively choose optimal viewpoints to better observe targets, TVDRNet employs a differentiable renderer to simulate this active process in 3D perception. By using text instructions as supervision to optimize intrinsic and extrinsic rendering parameters, the TVDRNet identifies the optimal viewpoints for observing the 3D scene, and therefore learning 'where to look' based on what the text instruction 'asked to find'. This process generates informative, task-relevant 2D images that are compatible with MLLMs. TVDRNet comprises: (1) the AVPL module, establishing a learnable mapping from semantics to optimal rendering parameters; and (2) the MGL module, fusing multi-modalities via semantic grouping to guide mask generation. Experiments show TVDRNet achieves SOTA performance in 3D reasoning segmentation (Reason3D, Instruct3D) and 3D visual grounding (ScanRefer).

Theory · Online Learning and Bandits

Taihei Oki, Shinsaku Sakaue

We study online inverse linear optimization, also known as contextual recommendation, where a *learner* sequentially infers an *agent*’s hidden objective vector from observed optimal actions over feasible sets that change over time. The learner aims to recommend actions that perform well under the agent’s true objective, and the performance is measured by the *regret*, defined as the cumulative gap between the agent’s optimal values and those achieved by the learner's recommended actions. Prior work has established a regret bound of $O(d\log T)$, as well as a finite but exponentially large bound of $\exp(O(d\log d))$, where $d$ is the dimension of the optimization problem and $T$ is the time horizon, while a regret lower bound of $\Omega(d)$ is known (Gollapudi et al. 2021; Sakaue et al. 2025). Whether a finite regret bound polynomial in $d$ is achievable or not has remained an open question. We partially resolve this by showing that when the feasible sets are *M-convex*—a broad class that includes matroids—a finite regret bound of $O(d\log d)$ is possible. We achieve this by combining a structural characterization of optimal solutions on M-convex sets with a geometric volume argument. Moreover, we extend our approach to adversarially corrupted feedback in up to $C$ rounds. We obtain a regret bound of $O((C+1)d\log d)$ without prior knowledge of $C$, by monitoring directed graphs induced by the observed feedback to detect corruptions adaptively.

Applications · Computer Vision

Yu Cao, Ziquan Liu, Zhensong Zhang, Jiankang Deng, Shaogang Gong, Jifei Song

Large-scale pre-trained video generators offer powerful priors for Video Super-Resolution (VSR), yet adapting them remains computationally prohibitive. Full fine-tuning demands extensive resources, and ControlNet-style adapters lose their efficiency advantage under modern Diffusion Transformers (DiTs) since the absence of encoder-decoder hierarchy forces duplication of the entire backbone. We observe that low-quality videos, despite degradation, retain reliable structural information such as layout and motion, and that such structural content is largely domain-agnostic. This suggests that a frozen generator can perform VSR when the input structure is properly aligned to its embedding space. Building on this insight, we propose LiteVSR, a minimalist framework that performs VSR using a completely frozen Diffusion Transformer (DiT) with a lightweight State-Aware Adapter. The adapter employs a dual-stream architecture that jointly processes static structural cues from the low-quality input and dynamic cues from intermediate denoising states through time-dependent cross-attention, enabling adaptive transition from structural alignment to texture refinement as denoising proceeds. LiteVSR achieves comparable restoration quality with only 12.68\% trainable parameters and 12 GPU-hours of training on a single A100.

Deep Learning · Robustness

Aymen Bouguerra, Daniel Vasquez, Alexandra Gomez-Villa, Chokri Mraidha, Fabio Arnez

Vision-Language Models (VLMs) such as CLIP have revolutionized zero-shot classification and safety-critical tasks, including Out-of-Distribution (OOD) detection. However, their high computational cost hinders efficient real-world deployment. While quantization is a standard solution for efficiency, its broader impact on reliability metrics beyond simple Top-1 accuracy remains critically under-explored. In this study, we conduct a large-scale evaluation of VLM quantization across a comprehensive experimental suite of over 700k evaluation runs with varying configurations. We find that, contrary to the assumption that quantization's noise degrades performance, it can simultaneously improve accuracy, calibration, OOD detection, and robustness to noise, though not to covariate shift or spurious correlations. We leverage these counterintuitive findings to characterize the mechanics of quantization beyond simple regularization: we show that quantization dampens high-rank spectral components, compelling the model to rely more heavily on robust, low-rank features. Ultimately, this spectral filtering effect drives the observed improvements in generalization and noise tolerance, establishing a pathway to deploy faster, more reliable VLMs by utilizing quantization beyond its conventional role.

Andrew Gordon, Daniel Murfet, William Snell, Garrett Baker, George Wang, Stan van Wingerden

Spectroscopy infers the internal structure of physical systems by measuring their response to perturbations. We apply this principle to neural networks: perturbing the data distribution by upweighting a token $y$ in context $x$, we measure the model's response via susceptibilities $\chi_{xy}$, which are covariances between component-level observables and the perturbation computed over a localized Gibbs posterior via stochastic gradient Langevin dynamics (SGLD). Theoretically, we show that susceptibilities decompose as a sum over \emph{modes} of the data distribution, explaining why tokens that follow their contexts ``for similar reasons'' cluster together in susceptibility space. Empirically, we apply this methodology to Pythia-14M, developing a conductance-based clustering algorithm that identifies 510 interpretable clusters ranging from grammatical patterns to code structure to mathematical notation. Comparing to sparse autoencoders, 50\% of our clusters match SAE features, validating that both methods recover similar structure.

Deep Learning · Large Language Models

Entang Wang, Yiwei Wang, Aleksandra Bakalova, Michael Hahn

In-context learning (ICL) excels at new tasks from minimal examples, yet we still lack a mechanistic explanation of how few-shot prompts shape a model’s function vector (FV)--a causal activation direction that drives task behavior on the ICL query. Across tasks and models, an $n$-shot FV is well-approximated by a linear combination of example-level sub-FVs, suggesting additive and composable contributions from individual demonstrations. Beyond additivity, we show that models contextualize individual examples' representations based on prior examples to adaptively reweight which demonstrations dominate the FV: attention shifts toward examples that are more informative and less ambiguous under the context. Finally, a causal decomposition separates Query–Key routing from Value updates, finding that contextualization’s most consistent contributions to FV quality arise from Query–Key alignment--particularly in ambiguous settings--while Value-mediated effects are more heterogeneous. Together, these results unify additive superposition with context-dependent attention reweighting into a mechanistic, testable account of how few-shot prompts implement tasks.

Benyu Zhang, Qiang Zhang, Jianpeng Cheng, Hong-You Chen, Qifei wang, Wei Sun, Shen Li, Jia Li, Jiahao Wu, Xiangjun Fan 等

Large Language Models (LLMs) represent a promising frontier for recommender systems, yet their development has been impeded by the absence of predictable scaling laws, which are crucial for guiding research and optimizing resource allocation. We hypothesize that this may be attributed to the inherent noise, bias, and incompleteness of raw user interaction data in prior continual pre-training (CPT) efforts. This paper introduces a novel, layered framework for generating high-quality synthetic data that circumvents such issues by creating a curated, pedagogical curriculum for the LLM. We provide powerful, direct evidence for the utility of our curriculum by showing that standard sequential models trained on our principled synthetic data significantly outperform ($+130\%$ on recall@100 for SasRec) models trained on real data in downstream ranking tasks, demonstrating its superiority for learning generalizable user preference patterns. Building on this, we empirically demonstrate, for the first time, robust power-law scaling for an LLM that is continually pre-trained on our high-quality, recommendation-specific data. Our experiments reveal consistent and predictable perplexity reduction across multiple synthetic data modalities. These findings establish a foundational methodology for reliable scaling LLM capabilities in the recommendation domain, thereby shifting the research focus from mitigating data deficiencies to leveraging high-quality, structured information.

Deep Learning · Large Language Models

Liew Yee Zhing, Andrew Tan, Anwar Majeed

Traditional hallucination detection fails on "Stubborn Hallucinations"—errors where LLMs are confidently wrong. We propose a geometric solution: Embedding-Perturbed Gradient Sensitivity (EPGS). We hypothesize that while robust facts reside in flat minima, stubborn hallucinations sit in sharp minima, supported by brittle memorization. EPGS detects this sharpness by perturbing input embeddings with Gaussian noise and measuring the resulting spike in gradient magnitude. This acts as an efficient proxy for the Hessian spectrum, differentiating stable knowledge from unstable memorization. Our experiments show that EPGS significantly outperforms entropy-based and representation-based baselines, providing a robust signal for detecting high-confidence factual errors.

Deep Learning · Robustness

Yue Liang, JIATONG DU, Ziyi Yang, Yanjun Huang, Hong Chen

Scene graphs provide structured abstractions for scene understanding, yet they often overfit to spurious correlations, severely hindering out-of-distribution generalization. To address this limitation, we propose CURVE, a causality-inspired framework that integrates variational uncertainty modeling with uncertainty-guided structural regularization to suppress high-variance, environment-specific relations. Specifically, we apply prototype-conditioned debiasing to disentangle invariant interaction dynamics from environment-dependent variations, promoting a sparse and domain-stable topology. Empirically, we evaluate CURVE in zero-shot transfer and low-data sim-to-real adaptation, verifying its ability to learn domain-stable sparse topologies and provide reliable uncertainty estimates to support risk prediction under distribution shifts.

Applications · Computer Vision

Muquan Li, Yingyi Ma, Yihong Huang, Hang Gou, KE QIN, Ming Li, Yuan-Fang Li, Tao He

Dataset distillation (DD) compresses a large training set into a small synthetic set for efficient training, but most DD methods optimize only clean accuracy and leave robustness uncontrolled. Recent robust DD methods improve robustness, yet they often suffer from a poor accuracy–robustness trade-off because they (i) treat all adversarially perturbed examples uniformly, despite robust risk being dominated by near-zero robust margins, and (ii) do not explicitly increase inter-class separation in the decision boundary where attacks concentrate. We present Contrastive Curriculum for Robust Dataset Distillation (C$^2$R), a margin-centric framework that couples an attack-aware curriculum with a contrastive robustness objective. From a robust-margin perspective, we derive a perturbation score that approximates each sample’s robust hinge, enabling a curriculum that prioritizes the smallest-margin adversaries that most directly drive robust error. In parallel, a class-balanced contrastive robustness loss enforces adversarial invariance while explicitly widening boundary separation across classes. Experiments on CIFAR-10/100, Tiny-ImageNet, and multiple ImageNet-1K subsets under six attacks show that C$^2$R achieves the best robust accuracy, outperforming prior robust DD methods by $2.8$\% on average. Under PGD, C$^2$R also reduces the average drop rate (DR) below $66.8$\% across datasets, indicating a stronger accuracy–robustness balance.

Reinforcement Learning · Everything Else

Changmin Yu, Máté Lengyel

The successor representation (SR) provides a powerful framework for decoupling predictive dynamics from rewards, enabling rapid generalisation across reward configurations. However, the classical SR is limited by its inherent policy dependence: policies change due to ongoing learning, environmental non-stationarities, and changes in task demands, making established predictive representations obsolete. Furthermore, in topologically complex environments, SRs suffer from spectral diffusion, leading to dense and overlapping features that scale poorly. Here we propose the Hierarchical Successor Representation (HSR) for overcoming these limitations. By incorporating temporal abstractions into the construction of predictive representations, HSR learns stable state features which are robust to task-induced policy changes. Applying non-negative matrix factorisation (NMF) to the HSR yields a sparse, low-rank state representation that facilitates highly sample-efficient transfer to novel tasks in multi-compartmental environments. Further analysis reveals that HSR-NMF discovers interpretable topological structures, providing a policy-agnostic hierarchical map that effectively bridges model-free optimality and model-based flexibility. Beyond providing a useful basis for task-transfer, we show that HSR's temporally extended predictive structure can also be leveraged to drive efficient exploration, effectively scaling to large, procedurally generated environments.

Reinforcement Learning · Everything Else

Mingxi Hu, Meiling Yu

Multi-objective reinforcement learning (MORL) must often support preferences that change online or are specified only after data collection. We study finite-horizon MORL with vector feedback in linear MDPs under two protocols: (i) predictable adversarial preferences revealed before each episode, and (ii) reward-free preference-free exploration (PFE), where exploration observes only transitions and must later answer arbitrary preference queries. Standard reductions are protocol-unsafe: re-scalarizing past stochastic rewards with future weights breaks the martingale structure needed for self-normalized confidence bounds, and hypervolume evaluation must account for episode-start randomization, which yields a deployable convex hull of return vectors. We propose a protocol-safe reward interface that estimates each reward coordinate via regression and performs scalarization only at query time, and we formalize deployable hypervolume semantics with a stability chain from support-function error to hypervolume error. Consequently, we obtain filtration-safe regret bounds for any predictable preference sequence without discretizing the simplex (only $\log m$ dependence) and matching near-minimax rates in linear MDPs, as well as sharp reward-free PFE guarantees: a (near-)minimax decision-optimal query answering rate $\tilde{O}(d^2 U_{\mathrm{ret}}^2/\varepsilon^2)$ and a tight separation from explicit transition-model recovery $\Theta(d(|\mathcal{S}|-1)/\varepsilon_P^2)$. These results connect online learning, preference-free deployment, and hypervolume-aware evaluation through a single protocol-aligned theory.

Reinforcement Learning · Policy Search

Xiangxiang Zhang, Caijun jia, Siyuan Li, he dingyu, Xiya Xiong, Zheng Sun, Honghao He, Yuchen Wu, Bihui Yu, Linzhuang Sun 等

Solving complex geometric problems inherently requires \textit{interleaved reasoning}: a tight alternation between constructing diagrams and performing logical deductions. Although recent Multimodal Large Language Models (MLLMs) have demonstrated strong capabilities in visual generation and plotting, we identify a counter-intuitive and underexplored phenomenon. Naively applying Supervised Fine-Tuning (SFT) on interleaved plot–solution data leads to a substantial degradation in reasoning performance compared to text-only baselines. We argue that this failure stems from a fundamental limitation of SFT, which primarily induces \textit{distributional alignment}: the model learns to reproduce the surface format of interleaved plotting but fails to internalize the causal dependency between the generated plot and reasoning steps. To overcome this limitation, we propose Faire (\textbf{F}unctional \textbf{a}lignment for \textbf{i}nterleaved \textbf{re}asoning), a reinforcement learning framework that enforces three casual constraints to move beyond superficial imitation toward \textit{functional alignment}. Extensive experiments show that Faire induces a qualitative shift in model behavior in which the plotting is effectively internalized, yielding competitive performance on challenging geometric reasoning benchmarks.

Applications · Robotics

Haldun Balim, Na Li, Yilun Du

Offline decision-making via diffusion models often produces trajectories that are misaligned with system dynamics, limiting their reliability for control. We propose *Model Predictive Diffuser* (MPDiffuser), a compositional diffusion framework that combines a diffusion planner with a dynamics diffusion model to generate task-aligned and dynamically plausible trajectories. MPDiffuser interleaves planner and dynamics updates during sampling, progressively correcting feasibility while preserving task intent. A lightweight ranking module then selects trajectories that best satisfy task objectives. The compositional design improves sample efficiency and adaptability by enabling the dynamics model to leverage diverse and previously unseen data independently of the planner. Empirically, we demonstrate consistent improvements over prior diffusion-based methods on unconstrained (D4RL) and constrained (DSRL) benchmarks, and validate practicality through deployment on a real quadrupedal robot.

Deep Learning · Everything Else

Tal Shuster, Eliya Nachmani

Recent neural audio codecs have achieved impressive reconstruction quality, typically relying on quantization methods such as Residual Vector Quantization (RVQ), Vector Quantization (VQ) and Finite Scalar Quantization (FSQ). However, these quantization techniques limit the geometric structure of the latent space, make it harder to capture correlations between features leading to inefficiency in representation learning, codebook utilization and token rate. In this paper we introduce Two-Dimensional Quantization (Q2D2), a quantization scheme in which feature pairs are projected onto structured 2D grids, such as hexagonal, rhombic, or rectangular tiling and quantized to the nearest grid values, yielding an implicit codebook defined by the product of grid levels, with codebook sizes comparable to conventional methods. Despite its simple geometric formulation, Q2D2 improves audio compression efficiency, with low token rates and high codebook utilization while maintaining state of the art reconstruction quality. Specifically, Q2D2 achieves competitive to superior performance in various objective and subjective reconstruction metrics, across extensive experiments in speech domain compared to state of the art models. Comprehensive ablation studies further confirm the effectiveness of our design choices.

Deep Learning · Attention Mechanisms

Biao Qian, Yang Wang, Yong Wu, Jungong Han

Data-Free Quantization (DFQ) addresses data security concerns by synthesizing fake samples, without accessing real data. It has garnered increasing attention in the context of Vision Transformers (ViTs), owing to the superiority of the self-attention mechanism compared to classical convolutional operation. However, previous DFQ arts for ViTs often suffer from a distribution mismatch between synthetic samples and input distribution expected by quantized models $Q$, resulting into the suboptimal performance. In this paper, we propose a novel Masked Attention Alignment approach for Data-Free Quantization of ViTs, named MaskAQ, revealing that: 1) the semantics in the self-attention mechanism is predominantly localized to a sparse subset of patches, called informative regions; 2) the informative regions dominate the mutual information between synthetic samples and $Q$'s outputs. To these ends, we incorporate differential entropy maximum over patch similarity of synthetic samples, which decouples informative regions from noisy background. To couple with varied $Q$, the informative regions are picked out to align full-precision models with $Q$ via a masked attention alignment objective, thus yielding high-quality synthetic samples. To further preserve mutual information between synthetic samples and updating $Q$, a periodic sample refreshing strategy comes up to endow MaskAQ with the capacity to continually adapt to the evolving state of $Q$ throughout the training process. Extensive experiments verify the merits of MaskAQ over state-of-the-art approaches across multiple backbones and downstream tasks, with Top-1 accuracy gain of up to 3.1% on ImageNet. Our code is available in supplementary material package.

Deep Learning · Large Language Models

Ariana Azarbal, Victor Gillioz, Vladimir Ivanov, Bryce Woodworth, jacob drori, Nevan Wichers, Aram Ebtekar, Alex Cloud, Alexander Turner

Developers often struggle to specify correct training labels and rewards. Perhaps they don't need to. We propose recontextualization, which reduces how often language models "game" training signals, performing misbehaviors those signals fail to penalize. We show recontextualization prevents models from learning to 1) prioritize evaluation metrics over chat response quality; 2) special-case code to pass incorrect tests; 3) overwrite evaluation functions rather than write correct code; and 4) become sycophantic. Our method works by generating completions from prompts discouraging misbehavior and then recontextualizing them as though they were in response to prompts permitting misbehavior. Recontextualization trains language models to resist misbehavior even when instructions permit it. This mitigates the reinforcement of misbehavior from misspecified training signals, reducing specification gaming without improving the supervision signal.