论文检索

Applications · Everything Else

Luankang Zhang, Hao Wang, Zhongzhou Liu, MINGJIA YIN, Yonghao Huang, Jiaqi Li, Wei Guo, Yong Liu, Huifeng Guo, Defu Lian 等

The scarcity of high-quality training data presents a fundamental bottleneck to scaling machine learning models. This challenge is particularly acute in recommendation systems, where extreme sparsity in user interactions leads to rugged optimization landscapes and poor generalization. We propose the Recursive Self-Improving Recommendation (RSIR) framework, a paradigm in which a model bootstraps its own performance without reliance on external data or teacher models. RSIR operates in a closed loop: the current model generates plausible user interaction sequences, a fidelity-based quality control mechanism filters them for consistency with user’s approximate preference manifold, and a successor model is augmented on the enriched dataset. Our theoretical analysis shows that RSIR acts as a data-driven implicit regularizer, smoothing the optimization landscape and guiding models toward more robust solutions. Empirically, RSIR yields consistent, cumulative gains across multiple benchmarks and architectures. Notably, even smaller models benefit, and weak models can generate effective training curricula for stronger ones. These results demonstrate that recursive self-improvement is a general, model-agnostic approach to overcoming data sparsity, suggesting a scalable path forward for recommender systems and beyond. Our anonymized code is available at https://anonymous.4open.science/r/RSIR-7C5B.

Deep Learning · Theory

Laura Ying Schulz, Daniel Mitropolsky, Tomaso A Poggio

While large models achieve impressive results, their learning dynamics are far from understood. Many domains of interest -- such as natural language syntax, coding languages, arithmetic problems -- are captured by context-free grammars (CFGs). In this work, we extend prior work on neural language modeling of CFGs in a novel direction: how language modeling behaves with respect to CFG substructure, namely "subgrammars". We first define subgrammars, and prove a set of fundamental theorems regarding language modeling and subgrammars. We show that language modeling loss (or equivalently the Kullback-Leibler divergence) recurses linearly over its top-level subgrammars; applied recursively, the loss decomposes into losses for "irreducible" subgrammars. We also prove that the constant in this linear recurrence is a function of the expected "recursion", a notion we introduce. We show that under additional assumptions, parametrized models learn subgrammars in parallel. Empirically, we confirm that small transformers learn subgrammars in parallel, unlike children, who first master simple substructures. We also briefly explore several other questions regarding subgrammars. We find that subgrammar pretraining can improve final performance, but only for tiny models relative to the grammar, while alignment analyses show that pretraining consistently lead to internal representations that better reflect the grammar’s substructure in all cases; we also observe persistent difficulty with deeper recursion, a limitation that appears even of large language models.

Haotian Chi, Zeyu Feng, Xingrui Yu, Linbo Luo, Yew Soon ONG, Ivor Tsang, Hechang Chen, Yi Chang, Haiyan Yin

Planning collaboration strategies for multi-agent embodied systems remains a core challenge for LLM-based planners, which often fail to capture the physical and coordination constraints of realworld environments. To address this, we present EvoCF, an agentic memory-driven evolutionary counterfactual planning framework for discovering improved multi-agent collaboration strategies through counterfactual plan generation and evaluation. First, we propose a symbolic constraint inductor that induces reusable symbolic constraints from failures, forming an evolving rule library. Then, we propose an evolutionary counterfactual plan generator that systematically explores semantically consistent plan variants through rule-conditioned mutations, enabling robust collaboration strategies beyond short-sighted one-shot LLM plans. Finally, we design an agentic memory-grounded evaluator that ranks candidate plans using retrieval-augmented evidence, producing interpretable, constraint-aware selections. Across multi-agent embodied simulation benchmarks, EvoCF consistently discovers more robust and executable plans compared to baseline approaches. Our results demonstrate that grounding multi-agent planning in agentic memory and counterfactual reasoning significantly enhances both effectiveness and robustness.

Social Aspects · Security

Wei Wang, Zihao Guan, Xing Zhou, Yan Ding, Yusong Tan, Jie Yu, Bao Li

Deploying large language models (LLMs) on untrusted hardware entails a risk of weight extraction, which can lead to unauthorized replication and misuse of the model. A practical approach is to leverage Trusted Execution Environments (TEEs) and protect model security by obfuscating model weights. However, existing obfuscation schemes struggle to simultaneously provide strong security guarantees and high performance: schemes with security guarantees incur substantial overhead due to frequent TEE interactions, whereas schemes that achieve efficient inference are insecure. We propose SLIM, a secure inference framework that exploits the iterative structure of LLMs to let transformed representations cascade through consecutive obfuscated layers, thereby minimizing interactions with the TEE. SLIM introduces a T-Way Mixing algorithm that performs consecutive inter-vector covering using carefully constructed block-diagonal Householder matrices and combines it with successive random permutations, providing thorough weight obfuscation while keeping TEE-side computation lightweight. Evaluations demonstrate that SLIM provides robust security guarantees and significantly outperforms prior state-of-the-art obfuscation schemes in terms of performance, delivering up to a $13.80\times$ speedup while preserving fidelity.

Applications · Everything Else

Ishaan Singh Chandok, Core Francisco Park

Scientific data annotation, such as tracking animals in video or proofreading neural reconstructions, remains bottlenecked by the “last mile” problem: even with strong automation, verification and correction consume substantial human effort. Standard approaches train models to directly predict annotations, discarding the rich supervision in how experts navigate, click, verify, and correct. We introduce a framework for studying behavioral cloning on scientific annotation: 9 synthetic tasks paired with synthetic annotations that simulate realistic human strategies including exploration, mistake correction, and strategic decision-making. Our experiments reveal several findings. First, skills emerge hierarchically: models learn GUI mechanics before task-critical decisions, and commit fewer mistakes than the training data while retaining the ability to correct errors when they occur. Second, scaling models on multi-task behavioral cloning shows that larger models are more data efficient, but exhibit worse decision-making despite similar placement accuracy. Third, multi-task pretraining enables efficient fine-tuning to new tasks, while training from scratch fails entirely. Fourth, linear probes reveal that models internally represent latent variables of the annotation process such as task phase and data position; interestingly, we find a shared mistake representation that generalizes across different annotation tasks. Overall, our framework establishes systematic benchmarks and identifies key bottlenecks, providing a foundation for scaling behavioral cloning to real-world scientific data annotation.

Deep Learning · Large Language Models

Haolong Qian, Xianliang Yang, Ma yinuo, Lirong Che, Feng Lu, Ye Guo, Lei Song, Jiang Bian, Chun Yuan

Knowledge distillation from powerful reasoning models underpins the development of Small Language Models (SLMs). A prevailing assumption in this paradigm is that training data with higher perceived quality, often defined by rigorous logic and superior reward scores, monotonically enhances downstream performance. In this paper, we identify a counter-intuitive \textbf{Quality-Utility Paradox} across diverse model families(Qwen2.5, LLaMA-3, DeepSeek): data refined by a superior Synthesis Oracle consistently underperforms the SLM's self-generated Rejection Sampling (RFT) data, despite achieving higher reward scores. We argue that Oracle models introduce an intrinsic representation bias that shifts training data into a distribution incompatible with the target SLM, where the SLM allocates limited computational capacity to stylistic imitation rather than logical reasoning. We utilize a \textbf{Style-Aligned Refinement} strategy to correct logical errors and strictly preserve the SLM's native syntax. Our experiments demonstrate that maintaining native syntax effectively mitigates syntactic adaptation costs, enabling distilled models to match or even surpass self-generated baselines. These findings underscore the necessity of syntactic alignment and advocate for model-aware reward designs that prioritize distributional compatibility alongside logical rigor. Our datasets and code will be publicly available.

Applications · Everything Else

Shiva Malay, Perampalli Shravan Nayak, Sagar Davasam, Srinivas Sunkara, Sai Rajeswar Mudumba

The rapid evolution of Large Language Models (LLMs) has shifted their role from passive information providers to active agents capable of executing complex workflows. However, the realization of a true "AI worker" is currently hindered by benchmarks that fail to capture the intricacy of professional environments, which demand long-horizon planning, complex tool usage, and adherence to strict access protocols. To bridge this gap, we introduce EnterpriseOps-Gym, a benchmark environment designed to evaluate agentic planning in realistic enterprise settings. EnterpriseOps-Gym provides: (i) 1,150 expert-curated tasks across eight interconnected domains (including HR, IT, Customer Service and productivity tools) that require managing persistent state and adhering to strict outcome-based verification logic; and (ii) a high-fidelity, containerized sandbox environment hosting 164 database tables and 512 functional tools. Our evaluation reveals critical limitations in state-of-the-art models: even the top-performing Claude Sonnet~4.5 achieves only 34.1\% success, struggling significantly with planning consistency, error recovery, and policy constraints. Furthermore, we observe that agents frequently fail to refuse infeasible tasks, leading to unintended and potentially harmful side effects on the system. These findings indicate that current agents are not yet ready for enterprise deployment. By releasing EnterpriseOps-Gym, we provide a concrete testbed to advance the reliability of autonomous agents in professional workflows.

Deep Learning · Attention Mechanisms

Xiaojie Yu, Haibo Zhang, Jeremiah D. Deng, Lizhi Peng

The maximal coding rate reduction ($\text{MCR}^2$) objective is proposed for learning low-dimensional subspace representations and for principled deep model design, where layer structures are derived by unrolling its optimization steps. However, existing methods motivated by this objective do not fully adhere to design principles implied by the $\text{MCR}^2$ gradient, which weakens the principled and interpretable foundations of the resulting models. In this work, we introduce PACEAttention, a novel principled attention mechanism inspired by the \textit{geometric insight }of $\text{MCR}^2$. From the geometric perspective, gradient-based updates of $\text{MCR}^2$ move features along directions shaped by the underlying low-dimensional feature structure. Our method captures this structure by leveraging randomization to guide feature updates. This principled construction enables the resulting PACENet to exhibit enhanced interpretability, with different heads attending to distinct image regions and capturing \textit{fine-grained} structures under simple supervised training. Besides, two learnable weights in PACEAttention enable explicit regulation of the feature update dynamics, reflecting the relative contributions of different components across layers. Experiments demonstrate that PACEAttention achieves superior performance and more stable scalability than previous principled modules while remaining low complexity.

Theory · Learning Theory

Dmitry Yarotsky, Eugene Golikov, Yaroslav Gusev

We develop a general mathematical framework to analyze scaling regimes and derive explicit analytic solutions for gradient flow (GF) in large learning problems. Our key innovation is a formal power series expansion of the loss evolution, with coefficients encoded by diagrams akin to Feynman diagrams. We show that this expansion has a well-defined large-size limit that can be used to reveal different learning phases and, in some cases, to obtain explicit solutions of the nonlinear GF. We focus on learning Canonical Polyadic (CP) decompositions of high-order tensors, and show that this model has several distinct extreme lazy and rich GF regimes such as free evolution, NTK and under- and over-parameterized mean-field. We show that these regimes depend on the parameter scaling, tensor order, and symmetry of the model in a specific and subtle way. Moreover, we propose a general approach to summing the formal loss expansion by reducing it to a PDE; in a wide range of scenarios, it turns out to be 1st order and solvable by the method of characteristics. We observe a very good agreement of our theoretical predictions with experiment.

Applications · Everything Else

Linbin Tang, Jingyan You, Zilin Kang, Hanzhang Liu, Sophia Zhang, Zenan Li, Chenrui Cao, Liangcheng Song, Jiaao Wu, Xian Zhang 等

Recent formal reasoning systems achieve IMO-level performance, but create a fragmented landscape: algebra and number theory use Lean, while geometry relies on domain-specific languages with limited formal guarantees. This fragmentation increases the trusted computing base and hinders unified model development. Existing geometry-in-Lean efforts (LeanEuclid, LeanGeo) introduce custom axiom systems incompatible with standard \mathlib{}, and their small scale ($<$1K problems) prevents large-scale training. However, native \mathlib{} autoformalization of geometry poses unique challenges: explicating implicit diagrammatic assumptions (e.g., topological configuration and non-degeneracy)---unlike existing custom systems that defer validity checks to external solvers---and adapting to \mathlib{}'s small, rapidly-evolving geometric constructs. We present \method{}, a framework that addresses these challenges through a four-stage pipeline---constraint explication, configuration anchoring, formalization mapping, and iterative repair---to automatically formalize geometry in native \mathlib{}. We construct OMNI-Geometry (768 competition problems) and Numina-Geometry (177,597 problems), the largest geometry formalization dataset in Lean. Human evaluation shows 48.89\% TOP1 and 73.33\% TOP5 accuracy. Training Goedel v2 on our formalizations improves proof success from 13.6\% to 15.1\%, validating dataset quality for unified neural theorem proving.

Applications · Neuroscience, Cognitive Science

Michael Ibrahim, Hanqi Zhao, Eli Sennesh, Zhi Li, Anqi Wu, Jacob Yates, Chengrui Li, Hadi Vafaii

Poisson-distributed latent variable models are widely used in computational neuroscience, but differentiating through discrete stochastic samples remains challenging. Two approaches address this: *Exponential Arrival Time* (EAT) simulation and *Gumbel-SoftMax* (GSM) relaxation. We provide the first systematic comparison of these methods, along with practical guidance for practitioners. Our main technical contribution is a modification to the EAT method that theoretically guarantees an unbiased first moment (exactly matching the firing rate), and reduces second-moment bias. We evaluate these methods on their distributional fidelity, gradient quality, and performance on two tasks: (1) variational autoencoders with Poisson latents, and (2) partially observable generalized linear models, where latent neural connectivity must be inferred from observed spike trains. Across all metrics, our modified EAT method exhibits better overall performance (often comparable to exact gradients), and substantially higher robustness to hyperparameter choices. Together, our results clarify the trade-offs between these methods and offer concrete recommendations for practitioners working with Poisson latent variable models.

Reinforcement Learning · Deep RL

Kazuki Ota, Takayuki Osa, Motoki Omura, Tatsuya Harada

Two-player games such as board games have long been used as traditional benchmark for reinforcement learning. This work revisits a regularized policy optimization with reverse Kullback-Leibler divergence and entropy divergence and analyzes this combination on two-player zero-sum settings from theoretical and empirical perspectives. From a theoretical perspective, we investigate the stability of the policy update rule on two theoretical settings: game-theoretic normal-form games and finite-length games. We provide convergence guarantees and verify our theoretical results by numerical experiments on synthetic games. From an empirical perspective, we derive a practical model-free reinforcement learning algorithm based on the regularized policy optimization. We validate the efficiency in training of our algorithm through comprehensive experiments on five board games: Animal Shogi, Gardner Chess, Go, Hex, and Othello. The experimental results demonstrate that our agent achieves more efficient learning than existing methods across the environments.

Applications · Computer Vision

Bo Du, Xiaochen Ma, Xuekang Zhu, Zhe Yang, Chaoqun Niu, Jian liu, Ji-Zhe Zhou

Fake Image Detection (FID), aiming at unified detection across four image forensic subdomains, is critical in real-world forensic scenarios. Compared with ensemble approaches, monolithic FID models are theoretically more promising, but to date, consistently yield inferior performance in practice. In this work, by discovering the "heterogeneous phenomenon'', which is the intrinsic distinctness of artifacts across subdomains, we diagnose the cause of this underperformance for the first time: the collapse of the artifact feature space driven by such phenomenon. The core challenge for developing a practical monolithic FID model thus boils down to the "unified-yet-discriminative" reconstruction of the artifact feature space. To address this paradoxical challenge, we hypothesize that high-level semantics can serve as a structural prior for the reconstruction, and further propose Semantic-Induced Constrained Adaptation (SICA), the first monolithic FID paradigm. Extensive experiments on our $ \textit{OpenMMSec} $ dataset demonstrate that SICA outperforms 15 state-of-the-art methods and reconstructs the target unified-yet-discriminative artifact feature space in a near-orthogonal manner, thus firmly validating our hypothesis. The code and dataset will be made publicly available.

Social Aspects · Safety

Tongxi Wu, Jian Zhang, Yang Gao

Safety alignment in large language models (LLMs) and multimodal large language models (MLLMs) is commonly assumed to operate as a near-binary threshold mechanism. We challenge this assumption by revealing that safety behavior is governed by an \emph{instability region} where small perturbations induce stochastic refusal decisions rather than deterministic outcomes. We develop a multi-metric diagnostic framework combining external and internal signals to characterize this instability. Through systematic experiments, we identify a characteristic \emph{diagnostic signature}: inputs in unstable regimes exhibit elevated output uncertainty yet \emph{decreased} internal safety activation, a decoupling phenomenon that explains why detection-based defenses fail against sophisticated attacks. Building on this framework, we introduce \textbf{Furina}, a jailbreak attack that deliberately induces this signature through fragmented, scene-anchored prompts without model-specific optimization. Furina outperforms strong single-turn and multi-turn baselines on HarmBench and achieves competitive results on MM-SafetyBench, demonstrating that uncertainty amplification provides a principled and transferable mechanism for understanding safety vulnerabilities. Code and supplementary materials: \url{https://anonymous.4open.science/r/Furina_Jailbreak-EF7C}.

Applications · Everything Else

Jiaxing Zhao, Hongbin Xie, Yuzhen Lei, Xuan Song, Zhuoran Shi, Lianxin Li, Shuangxue Liu, Haoran Zhang

Multi-Agent Systems (MAS) powered by Large Language Models (LLMs) have recently become a strong paradigm for solving complex workflow-structured tasks through expert collaboration. However, the data that make such collaboration effective are typically distributed across organizations and cannot be centrally pooled due to privacy, intellectual property, and compliance constraints. Federated Learning preserves data locality, yet most federated paradigms treat clients as independent and fail to capture workflow dependencies that are essential for coherent multi-stage collaboration. Data locality and workflow dependency are orthogonal, and the key challenge arises where both must be satisfied, namely federated, workflow-aware collaboration. We introduce FedWave, a framework that enables LLM-based experts to solve sequential workflows under strict privacy constraints. FedWave integrates a Value Chain Layer that encodes inter-stage dependencies with communication-efficient federated LoRA adaptation, a server-side Mixture-of-Experts (MoE) router that performs input-conditioned expert fusion at inference time while retaining standard federated aggregation during training, and a Direct Preference Optimization (DPO) stage that aligns collaborative outputs using router-induced preferences. Experiments show that FedWave consistently outperforms strong federated baselines and remains competitive with centralized multi-agent systems without compromising data privacy. Code is available at https://anonymous.4open.science/r/FedWave-111A.

Deep Learning · Everything Else

Xin Yang, Yemin Wang, Mingda Liu, Letian Li, Shuaishuai Cao, ZhengXiao He, Ryan Dong

Scaling large language models (LLMs) has driven their success, yet dense Transformers couple capacity and computation: every parameter is activated for every token, making training and inference costs grow linearly with model size—a critical bottleneck as models approach trillion-parameter regimes. We aim to scale capacity through MoE-style mixture throughout the LLM pipeline rather than only the FFN. Prior pipeline-level approaches include ParaScale, which introduces virtual tokens and parallel streams but incurs substantial overhead and suffers from homogenized routing and gradient collapse, and AltUp, which uses an auxiliary prediction branch but offers limited adaptivity and slow convergence. We establish that MoE-style mixture layers can be reformulated as variable-kernel dynamic convolutions, where each expert corresponds to a $1{\times}1$ convolutional kernel and routing implements input-conditioned kernel aggregation. Building on this equivalence, we introduce cMoLLM: a convolutionally gated mixture-of-LLMs that routes over end-to-end streams through fully differentiable dynamic convolution. In GPT-2-style models trained on FineWeb, cMoLLM improves language modeling perplexity and downstream GLUE and SQuAD accuracy under matched compute, with better stream utilization, more stable optimization, and favorable scaling compared to ParaScale- and AltUp-style baselines.

Deep Learning · Large Language Models

Ido Amit, Ido Galil, Ran El-Yaniv

As LLMs generate increasingly long outputs, effective uncertainty estimation must identify errors at fine-grained levels rather than discard entire responses. While such methods exist, evaluating uncertainty at any resolution (token to an entire generation) is challenging and highly sensitive to label imperfections, making zero-noise benchmarks essential; yet, long-form generation benchmarks tend to rely on fallible labels rather than deterministic ground truth. We introduce Single-answer Atomic Long-form Target (SALT), a benchmark of six procedurally generated tasks with single deterministic long textual ground truths, enabling unit-level evaluation of correctness, calibration, and ranking without external judges. Equipped with SALT, our analysis of 50+ LLMs reveals key insights: We identify which confidence functions dominate each uncertainty aspect and show that effective ranking benefits more from coarser evaluation resolutions; SALT further facilitates precise calibration tracking throughout generation, revealing a divergence in the accuracy–calibration relationship, with high- and low-performing models exhibiting degradation ($\rho=0.87$) and improvement ($\rho=-0.92$). Finally, we demonstrate that reasoning, via Chain-of-Thought prompting or internalized through training, introduces a trade-off, improving accuracy while degrading confidence ranking. These findings directly impact risk-critical applications requiring reliable error identification and mitigation.

Applications · Everything Else

Yanfang Liu, Mingjun Wang, Peng XU, Rongliang Fu, Bei Yu, Tsung-Yi Ho

Analog circuits constitute the indispensable interface between physical reality and digital computation, underpinning safety-critical systems from autonomous driving to medical implants. Consequently, verification correctness is paramount; yet, it remains the critical bottleneck in hardware design, consuming over 50\% of engineering cycles due to a heavy reliance on the manual interpretation of unstructured, heterogeneous specifications. While Large Language Models (LLMs) offer automation potential, their probabilistic, autoregressive nature is structurally misaligned with the strict determinism required for analog verification tasks. Specifically, generic LLMs struggle to resolve semantic dispersion, latent causal dependencies, and numerical precision. To bridge this gap, we introduce AnalogVerifier, a neuro-symbolic framework that automates end-to-end testbench generation by decoupling semantic translation from logical enforcement. We propose a four-stage architecture: (1) Context-Aware Task Serialization transforms complex specifications into atomic tasks via an agentic workflow; (2) Graph-Symbolic Scheduling satisfies analog design constraints through Port Dependency Graphs (PDG) for correct-by-construction sequencing; (3) Numerical-Symbolic Grounding mitigates numerical hallucination by delegating threshold derivation to a deterministic symbolic oracle; (4) Closed-Loop Repair enables correctness and completeness of the generated testbenches by simulation feedback. Evaluation on five industrial analog circuits demonstrates that AnalogVerifier achieves 82.3\%--100\% functional pass rate, establishing a new paradigm for reliable, automated analog verification. The code and data are publicly available at \url{https://anonymous.4open.science/r/ICML26--AnalogVerifier-72EE/}.

Deep Learning · Large Language Models

Xinpeng Wang, William Cao, Andrew Wilson, Zhe Zeng

Recent studies on hallucination detection have shown that hallucination-related signals are more strongly encoded in intermediate layers than in the final layer of large language models (LLMs). While a growing body of work has sought to exploit this property for hallucination detection, the problem of how to automate the selection of high-performing layers is underexplored, and the development of principled methods for this purpose remains an open challenge. To address this gap, we first propose several hypotheses for why such signals emerge in intermediate layers and test corresponding criteria for automatic layer selection. We evaluate these criteria across two LLM architectures and five datasets, and find that none of them deliver satisfying performance. Instead, we propose a new selection criterion, First Effective Peak of Intrinsic Dimension (FEPoID), that is able to consistently identify optimal or near-optimal layers and outperforms the aforementioned criteria and existing hallucination detection baselines. This criterion is training-free and requires negligible computational overhead. Additionally, we study the generation behaviors of LLMs and introduce a simple yet effective truncation strategy, which further amplifies the hallucination-related signals and leads to substantial improvements in overall detection performance.

Deep Learning · Generative Models and Autoencoders

Julian Wustl, Philipp Haid, Yarema Okhrin, Claudius Schnörr

Codebook-based generators built on masked language model (MLM) transformers have become highly effective in text and vision, yet remain underused for tabular data. This is because codebooks typically act as information bottlenecks, whereas tabular generation requires them to generalize. We address this gap with Q-Tab, a codebook-based tabular generator based on lookup-free quantization (LFQ) with residual corruption. The resulting corruption kernel induces a moving Nadaraya–Watson–style kernel regression over a large discrete code space, which turns codebook learning into a moving-target problem. We derive necessary conditions for the learnability of such moving codebooks and show how the residual LFQ construction aligns with these conditions. Q-Tab achieves state-of-the-art downstream predictive utility and missing-value imputation, while matching the distributional fidelity of diffusion-based generators, notably without any post-hoc temperature tuning.