论文检索

Deep Learning · Everything Else

Hangfeng Liang, Yutao Hu, Yanhan Hu, Xiaohan Wu, WENQI SHAO, Ying Fu

Low-light video enhancement (LLVE) remains a challenging task due to severe information degradation under low-illumination conditions. Recent multimodal approaches have significantly improved enhancement performance by incorporating auxiliary modalities, such as event streams and infrared images. However, these methods typically assume the availability of these modalities at inference, which is often not feasible in real-world scenarios. To solve this problem, in this work, we propose AMNet, a unified multimodal framework for LLVE, to support flexible modality-agnostic inference, where auxiliary modalities may be unavailable. To address the issue of modality absence, we introduce a Spatial-Spectral Dual-Gated Translator that learns the correspondence between auxiliary modalities and RGB inputs, producing implicit auxiliary representations to support the robust enhancement. Additionally, to fully facilitate the learning of cross-modal correspondence, we conduct large-scale multimodal pretraining based on the RGB-only dataset with synthetic auxiliary modalities. Extensive experiments demonstrate that AMNet could handle arbitrary inference-time modality combinations and exhibits superior performance for LLVE under modality absence conditions.

Deep Learning · Everything Else

Xiao Zhang, Zengzhe Chen, Mingyi Li, Jing Qiao, Fuzhen Zhuang, YUAN YUAN, Dongxiao Yu

Decentralized class continual learning refers to a paradigm where distributed clients continuously acquire new classes while retaining previously learned information without relying on a central server. With increasing emphasis on privacy preservation, there is a growing need for on-demand unlearning, introducing two key challenges: Historical Class Unlearning and Network-Wide Knowledge Entanglement. In this work, we propose a decentralized continual learning framework with on-demand unlearning (DCU), which is the first attempt at achieving class continual learning and arbitrary-time class unlearning in a distributed setting. Specifically, our proposed DCU comprises three main stages: prototypes extraction, prototype-guided continual learning, and unlearning with disposable prototypes. Firstly, the prototypes extraction mechanism is designed to capture the class-specific concepts as lightweight, disposable embeddings. Then, the synthetic data guided by these prototypes can be combined with real data to achieve incremental learning through distillation. Besides, synthetic samples with noisy label are used to guide the adjustment of the model's decision boundary, effectively erasing the influence of the target class while preserving other classes' knowledge. Extensive experiments conducted on two datasets demonstrate the effectiveness of our DCU in dynamic learning and target class unlearning.

Deep Learning · Everything Else

Wenhai Wan, Teng Zhang, Shao-Yuan Li, Xinrui Wang, Qiang-Sheng Hua, Songcan Chen

Real-world datasets often follow a long-tailed distribution, making generalization to tail classes difficult. We revisit this problem through the lens of shortcut learning, where models prefer the easiest predictive cues (e.g., background or textures) over object-centric semantics, especially under scarce and biased supervision. We find that this tendency is amplified for tail classes: limited examples often share similar contexts, making non-semantic signals highly correlated and thus tempting shortcuts, whereas head classes with diverse appearances and environments encourage more stable object-focused representations. Motivated by this observation, we propose Shortcut-Resistant CAM Distillation (SRCD), a plug-and-play framework that transfers object-focused explanations from head to tail classes. SRCD operates in the Class Activation Map (CAM) space, where a CAM provides a class-specific spatial evidence map for a prediction. SRCD aggregates CAMs from a small set of head-class candidates into a shortcut-resistant teacher using an energy-model weighting based on coherence (Laplacian smoothness) and concentration (Hoyer sparsity), and distills it to the tail-class CAM. We provide a theoretical analysis that quantifies shortcut reliance as shortcut-region evidence mass in CAM space and shows that SRCD suppresses tail shortcuts. Extensive experiments on long-tailed benchmarks consistently improve strong baselines.

Deep Learning · Everything Else

Tianjun Yao, Zhaoyi Li, Zhiqiang Shen

Multi-agent systems (MAS) built on large language models (LLMs) have demonstrated remarkable performance across diverse tasks. Existing approaches optimize communication topology, role assignment, or LLM routing in isolation, while treating each agent as a monolithic unit—failing to exploit internal LLM mixtures that can enhance individual role capabilities. We propose **`HieraMAS`**, a hierarchical agent collaboration framework with intra-node LLM mixtures and inter-node communication topology. HieraAgent introduces *supernodes*, where each functional role comprises multiple heterogeneous LLMs in a propose-synthesis structure. The optimization of **`HieraMAS`** poses unique credit assignment challenges, as final task performance heavily depends on LLM capabilities, potentially causing erroneous reinforcement of suboptimal configurations. We address this via a two-stage algorithm: (1) multi-level reward attribution providing fine-grained feedback at both node and system levels; and (2) graph classification treating topology selection as a holistic task rather than per-edge optimization. Experiments on reasoning and coding benchmarks demonstrate that **`HieraMAS`** significantly outperforms existing methods while achieving better cost-performance trade-offs.

Deep Learning · Everything Else

Adir Dayan, Yam Eitan, Haggai Maron

Weight-space learning studies neural architectures that operate directly on the parameters of other neural networks. Motivated by the growing availability of pretrained models, recent work has demonstrated the effectiveness of weight-space networks across a wide range of tasks. SOTA weight-space networks rely on permutation-equivariant designs to improve generalization. However, this may negatively affect expressive power, warranting theoretical investigation. Importantly, unlike other structured domains, weight-space learning targets maps operating on both weight and function spaces, making expressivity analysis particularly subtle. While a few prior works provide partial expressivity results, a comprehensive characterization is still missing. In this work, we address this gap by developing a systematic theory for expressivity of weight-space networks. We first prove that all prominent permutation-equivariant networks are equivalent in expressive power. We then establish universality in both weight- and function-space settings under mild, natural assumptions on the input weights, and characterize the edge-case regimes where universality no longer holds. Together, these results provide a strong and unified foundation for the expressivity of weight-space networks.

Deep Learning · Everything Else

Ofir Gordon, Lior Dikstein, Arnon Netzer, Idan Achituve, Hai Victor Habi

Post-training quantization (PTQ) is a widely used approach for reducing the memory and compute costs of large language models (LLMs). Recent studies have shown that applying invertible transformations to activations can significantly improve quantization robustness by reducing activation outliers; however, existing approaches are largely restricted to rotation or Hadamard-based transformations. Moreover, most studies focused primarily on traditional quantization schemes, whereas modern hardware increasingly supports the microscaling (MX) data format. Attempts to combine both showed severe performance degradation, leading prior work to introduce assumptions on the transformations. In this work, we take a complementary perspective. First, we provide a theoretical analysis of transformations under MX quantization by deriving a bound on the quantization error. Our analysis emphasizes the importance of accounting for both the activation distribution and the underlying quantization structure. Building on this analysis, we propose LATMiX, a method that generalizes outlier reduction to learnable invertible affine transformations optimized using standard deep learning tools. Experiments show consistent improvements in average accuracy for MX low-bit quantization over strong baselines on a wide range of zero-shot benchmarks, across multiple model sizes.

Deep Learning · Everything Else

Riju Marwah, Ritvik Garimella, Vishal Pallagani, Atishay Jain, Michael Stewart, Amit Sheth

Autoregressive language models frequently degrade during long-horizon generation, producing repetitive text, losing instruction adherence, and exhibiting unstable entropy. Despite the prevalence of these failures, practitioners lack online diagnostics to detect them in real time as they occur. We formalize this degradation as cognitive fatigue, a measurable generation-time state characterized by decay in attention to the original prompt, representational drift, and entropy miscalibration. We introduce the Fatigue Index (FI), a lightweight, model-agnostic diagnostic that aggregates these three signals under explicit axioms (monotonicity, boundedness, interpretability), enabling reliable runtime monitoring. Across nine models (1B–13B parameters), FI trajectories exhibit structured temporal dynamics, predict task degradation (AUROC = 0.95) and repetition (ρ = 0.94), and reveal non-monotonic scaling behavior: instruction-tuned models below 3B exhibit faster collapse than base models, with this trend reversing at 7B. Stress analyses further show that FI onset accelerates under longer contexts, middle-positioned evidence, and reduced numerical precision. These results establish cognitive fatigue as a coherent and measurable phenomenon, and position FI as a principled tool for runtime reliability monitoring in production LLM systems.

Deep Learning · Everything Else

Marco Federici, Boris van Breugel, Paul Whatmough, Markus Nagel

Quantization can drastically increase the efficiency of large language and vision models, but typically incurs an accuracy drop. Recently, function-preserving transforms (e.g. rotations, Hadamard transform, channel-wise scaling) have been successfully applied to reduce post-training quantization error, yet a principled explanation remains elusive. We analyze linear-layer quantization via the signal-to-quantization-noise ratio (SQNR), showing that for uniform integer quantization at a fixed bit width, SQNR decomposes into (i) the concentration of weights and activations (capturing spread and outliers), and (ii) the alignment of their dominant variation directions. This provides an actionable insight: enhancing alignment between weight and activation variation directions can reduce quantization error, complementing concentration-focused approaches. Motivated by this, we introduce Concentration–Alignment Transforms (CAT), a lightweight linear transformation that uses a covariance estimate from a small calibration set to jointly improve concentration and alignment, approximately maximizing SQNR. Experiments across several LLMs show that CAT consistently matches or outperforms prior transform-based quantization methods at 4-bit precision.

Deep Learning · Everything Else

Thomas Shih-Chao Liang, Zhuoran Yu, Yong Jae Lee

Large Language Models (LLMs) possess broad conceptual knowledge acquired through large-scale text pretraining, yet their potential to supervise models in other modalities remains underexplored. In this work, we propose \LaViD—Language-to-Visual Knowledge Distillation—a simple and effective framework for transferring high-level semantic knowledge from a language-only teacher to a vision-only student model. Instead of relying on paired multimodal data, LaViD elicits conceptual signals from an LLM by prompting it to generate multiple-choice questions (MCQs) that probe semantic distinctions between visual classes. Each class is mapped to a soft label distribution over these MCQs, forming a rich conceptual signature that guides the student through an auxiliary distillation loss. Notably, despite using a language-only teacher without access to image data, LaViD consistently outperforms recent methods like MaKD that distill from vision-language models across multiple fine-grained benchmarks. It also achieves competitive or superior performance compared to state-of-the-art visual distillation methods such as DKD and MLKD, with further gains when combined with logit standardization. On the Waterbirds dataset, LaViD substantially improves worst-group accuracy, demonstrating enhanced robustness to spurious correlations with distillation.

Reinforcement Learning · Everything Else

Xinyu Zhang

Asynchronous reinforcement learning enables high-throughput training but introduces *policy lag*, where experiences are collected under stale policy weights. We identify a key phenomenon in code generation: **gradient variance scales exponentially with task difficulty under staleness**, because hard tasks have narrow solution spaces corresponding to sharp loss landscape curvature (high Hessian eigenvalues). We formalize this as a *staleness budget optimization problem* and prove that the optimal allocation follows an exponential decay: $\eta^*(d) = \eta_{base} \cdot e^{-\lambda d}$ where $\lambda = \alpha/2$ is half the Hessian growth rate. Building on this theory, we propose ACEAS (**A**daptive **C**urriculum with **E**xecution-**A**ware **A**sync **S**cheduling), combining bandit-based curriculum selection, execution-aware staleness budgets, and curriculum-staleness coupling derived from first principles. Our mechanistic analysis validates the theoretical predictions: the "safe zone" of gradient coherence follows the derived exponential boundary. On code generation benchmarks, ACEAS achieves over 2$\times$ higher throughput than synchronous baselines while improving Pass@1 from 39.7% to 60.1%, demonstrating that principled staleness control grounded in loss landscape geometry enables efficient asynchronous curriculum learning.

Applications · Everything Else

Youngsung Kim

Differentiable discrete selection uses soft mixtures during training but hard selection at deployment, resulting in a training-inference gap. We decompose this gap into selection gap (method-dependent, reducible) and computation gap (input-dependent, irreducible). Our key finding: the selection gap is determined by forward-pass structure, not backward-pass gradients. Methods using hard selection during training achieve zero selection gap by construction, while mixture methods exhibit gaps even with identical gradient estimators. This occurs because mixtures reward hedging across options, while deployment requires commitment to one. We propose CAGE (Confidence-Adaptive Gate Exploration), which addresses optimization entirely in the backward pass by adapting temperature based on selection confidence. We also identify a critical failure mode: Gumbel-ST suffers 40--50 percentage point accuracy collapse at low temperatures, which CAGE prevents. Experiments on logic gate networks validate the theory: hard selection achieves 98% accuracy with zero gap across all temperatures.

Social Aspects · Everything Else

Ujwal Kumar, Alice Saito, Hershraj Niranjani, Rayan Yessou, Tan Phan Xuan

Constitutional AI has focused on single-model alignment using fixed principles. However, multi-agent systems create novel alignment challenges through emergent social dynamics. We present Constitutional Evolution, a framework for automatically discovering behavioral norms in multi-agent LLM systems. Using a grid-world simulation with survival pressure, we study the tension between individual and collective welfare, quantified via a Societal Stability Score $\mathcal{S} \in [0,1]$ that combines productivity, survival, and conflict metrics. Adversarial constitutions lead to societal collapse ($\mathcal{S}=0$), while vague prosocial principles (''be helpful, harmless, honest'') produce inconsistent coordination ($\mathcal{S}=0.249$). Even constitutions designed by Claude 4.5 Opus with explicit knowledge of the objective achieve only moderate performance ($\mathcal{S}=0.332$). Using LLM-driven genetic programming with multi-island evolution, we evolve constitutions maximizing social welfare without explicit guidance toward cooperation. The evolved constitution $\mathcal{C}^*$ achieves $\mathcal{S}=0.556\pm0.008$ (123\% higher than human-designed baselines, $N=10$), eliminates conflict, and discovers that minimizing communication (0.9\% vs 62.2\% social actions) outperforms verbose coordination. Our interpretable rules demonstrate that cooperative norms can be discovered rather than prescribed.

Social Aspects · Everything Else

Rashid Mushkani

This position paper argues that prompts used to deploy large language models (LLMs) in public-sector settings should be treated as governed artefacts rather than private, transient inputs. Prompts encode role instructions, decision framings, and value claims; prompt choice can materially shift outputs even when model weights and input records are held fixed. Existing governance tools, including model and dataset documentation, organisation-level policies, and post-training alignment, rarely make the local prompt collections used in deployment transparent, contestable, or auditable. We propose Prompt Commons: a versioned, community-maintained repository of prompt templates with provenance metadata, licensing, and moderation logs. Using a pilot dataset collected with community partners in a large North American city (443 human prompts; 3,317 after augmentation), we illustrate three governance states (open, curated, veto-enabled) and a negotiation-oriented ensemble method that aggregates stakeholder prompts into compromise recommendations. We close with falsifiable implications and an evaluation agenda for prompt-layer governance.

Social Aspects · Everything Else

Wenjun Cao

This position paper argues that modern machine learning creates structural risks for knowledge and cultural production, operating before AGI thresholds through market selection mechanisms. We use \emph{temporality} operationally for how understanding changes over time and signals left by that process. Representation learning and autoregressive generation approximate output distributions while omitting slow, path-dependent human learning; at scale, these function as general-purpose production technologies. We analyze the link from technical indistinguishability to market selection: when divergence between model and temporal signals is small and verification costly, decision makers cease screening, prices track pooled quality, and temporality-intensive work exits. We call this phenomenon \emph{value collapse}. Recent evidence shows this active: academic publishing has experienced dramatic productivity increases alongside troubling quality trends; cultural production shows explosive AI-generated content adoption. As training data mirror such environments, models absorb their outputs and model collapse risk rises. Alignment is orthogonal: by narrowing observable gaps, it intensifies selection pressures where provenance remains costly.

Probabilistic Methods · Everything Else

Shi-ang Qi, Yakun Yu, Russell Greiner

Likelihood-based training is the dominant paradigm in survival prediction. Under independent censoring, we can factorize the likelihood and optimize only the terms related to event modeling, effectively treating the censoring mechanism as incidental. This is justified when censoring is *non-informative*, i.e., when the censoring process shares no parameters with the event-time model. However, this may not hold in practice, and ignoring censoring contributions may discard useful signals for learning representations that can help to effectively estimate event distributions. Motivated by this, we argue that explicitly modeling censoring can improve representation learning and time-to-event estimation, particularly when event and censoring processes are coupled. We introduce a latent decomposition view that partitions covariates into four disjoint factors: those affecting only the event process, only the censoring process, both, or neither. We then learn decomposed representations for the first three categories to guide a better estimation of the event distribution. We instantiate our method on four popular deep-learning survival models and evaluate on 10 datasets (2 semi-synthetic and 8 real-world), showing consistent gains over strong baselines and multiple SOTA methods.

General Machine Learning · Everything Else

Shutong Chen, Tianyi Zhou, Guodong Long, Jie Ma, Jing Jiang, Chengqi Zhang

Contemporary AI faces the challenge of balancing generality with user-specific personalization. In federated learning (FL), this challenge is amplified by highly heterogeneous client data with complex non-IID patterns beyond standard modeling assumptions. Many existing FL methods are designed for relatively restricted heterogeneity settings (e.g., a fixed number of clusters or a fixed form of personalization), limiting their robustness under complex structures. In this work, we study FL from a \emph{multi-level non-IID} perspective, where client similarity is approximated by multiple granularities of shared knowledge: global, subgroup, and client-specific components. This view captures coarse-to-fine relationships while requiring less prior knowledge of task boundaries. Building on this insight, we propose \emph{Federated Multi-level Additive Modeling} (FeMAM), which learns multiple levels of shareable models and constructs personalized predictors via additive composition across levels. To move beyond a fixed structure, FeMAM allows models to grow and be pruned dynamically during training, adapting to diverse federated scenarios. Despite employing multiple models, FeMAM remains cost-friendly by activating only a small subset (one level) of models for training at a time. Extensive experiments show that FeMAM effectively approximates complex non-IID structures and consistently outperforms representative clustered and personalized FL baselines.

General Machine Learning · Everything Else

Yipeng Lin, Fengqiang Wan, Yang Yang

Pre-trained models with parameter-efficient fine-tuning have shown strong effectiveness in Class-Incremental Learning (CIL), which seeks to balance model plasticity and stability. In this context, orthogonality constraints can significantly enhance model stability, yet their reliance on subspace inevitably compromises model plasticity over long tasks. To address this, we propose Gradient-Recycling Low-Rank Adaptation (GR-LoRA), which reconciles stability and plasticity by recycling the gradients discarded in orthogonal projection. Specifically, GR-LoRA recycles post-decomposition non-orthogonal gradient components into task-specific lightweight modules and selects optimal module via entropy to improve plasticity, while incorporating local and global mismatch suppression to preserve stability by synthesizing out-of-distribution representations across all tasks. Theoretical analysis confirms that this recycling strategy preserves stability and improves plasticity. Experimental results from multiple CIL benchmarks verify the effectiveness and general applicability of GR-LoRA.

Reinforcement Learning · Everything Else

Abhishek Vijaya Kumar, Bhaskar Kataria, Byungsoo Oh, Emaad Manzoor, Rachee Singh

In RL post-training of LLM agents, calls to ex ternal tools take several seconds or even min- utes, leaving allocated GPUs idle and inflating post-training time and cost. While many tool invocations repeat across parallel rollouts and could in principle be cached, naively caching their outputs for reuse is incorrect since tool outputs depend on the environment state induced by prior agent interactions. We present TVCACHE, a stateful tool-value cache for LLM agent post- training. TVCACHE maintains a tree of observed tool-call sequences and performs longest-prefix matching for cache lookups: a hit occurs only when the agent’s full tool history matches a previ- ously executed sequence, guaranteeing identical environment state. On three diverse workloads— terminal-based tasks, SQL generation, and video understanding—TVCACHE achieves cache hit rates of up to 70% and reduces median tool call execution time by up to 6.9×, with no degradation in post-training reward accumulation.

Applications · Everything Else

chunyang yu, Ning SUN, Shenyue Wang

Gesture interaction and touchless sensing is a natural and intuitive control method that allows users to control devices through natural hand or body movements, reducing reliance on physical input and enhancing convenience and functional efficiency. The biggest challenge is balancing accuracy and model parameter count for custom gesture tasks, while minimizing the number of data samples required for training. This paper proposes a novel IMU-based BiCrossNet model and two novel data augmentation models (delta-generator and embedding-generator) to address this challenge. Compared with existing methods, the model proposed in this article boosts accuracy by 11.7% and 12.7% in UMAHand (public datasets); and 8.85% and 5.25% in GRHand (self-developed datasets), with decreasing 27.8% pretrained feature-extractor model parameters. This research lays a solid foundation for deploying and implementing custome gesture recognition engineering on intelligent terminal devices.

Theory · Everything Else

Zachary Izzo, Iain Melvin

We study the problem of subgroup discovery for survival analysis, where the goal is to find an interpretable subset of the data on which a Cox model is highly accurate. We examine why existing quality functions are insufficient for this problem and introduce two technical innovations: the *expected prediction entropy (EPE)*, a novel metric for evaluating survival models that predict hazard functions, and the *conditional rank statistics (CRS)*, which quantifies individual point deviation from a subgroup's survival time distribution. We study the EPE and CRS theoretically and show they address problems with existing metrics. We then introduce seven algorithms for Cox subgroup discover. Our main algorithm is based on the DDGroup framework of Izzo et al. (2023) and leverages both the EPE and CRS, allowing theoretical correctness guarantees in well-specified settings. Empirical evaluation on synthetic and real data confirms our theory, showing our methods recover ground-truth subgroups in well-specified cases and achieve better model fit than naively fitting the Cox model to the entire dataset. A case study on NASA jet engine simulation data demonstrates that discovered subgroups uncover known nonlinearities in the data and suggest design choices mirrored in practice.