arXivSub Start free trial

NeurIPS 2025 Papers with Code β€” Page 4

Conference on Neural Information Processing Systems Β· 2283 papers

Class conditional conformal prediction for multiple inputs by p-value aggregation

Jean-Baptiste Fermanian (University of Montpellier), Joseph Salmon (University of Montpellier)

CodeClassificationRecognitionConvolutional Neural NetworkImage

🎯 What it does: A category conditional conformal prediction method is proposed for instances with multiple images, utilizing the p-values generated from multiple images for aggregation to construct smaller and more reliable prediction sets.

Class-aware Domain Knowledge Fusion and Fission for Continual Test-Time Adaptation

Jiahuan Zhou (Peking University), Gang Hua (Amazon.com, Inc)

CodeClassificationDomain AdaptationTransformerPrompt EngineeringImage

🎯 What it does: For Continuous Testing Task Adaptation (CTTA), a Category-Aware Knowledge Fusion and Splitting (KFF) framework is proposed to dynamically learn and accumulate knowledge from different domains while avoiding catastrophic forgetting.

CLAWS:Creativity detection for LLM-generated solutions using Attention Window of Sections

Keuntae Kim (Hanyang University), Yong Suk Choi (Hanyang University)

CodeClassificationComputational EfficiencyTransformerLarge Language ModelSupervised Fine-TuningPrompt EngineeringText

🎯 What it does: This paper proposes a detection method called CLAWS, which uses the attention distribution of various parts of the prompt to determine whether the mathematical answers generated by LLM are Hallucinated, Creative, or Typical, achieving three-class classification without the need for manual evaluation.

Clip-and-Verify: Linear Constraint-Driven Domain Clipping for Accelerating Neural Network Verification

Duo Zhou (University of Illinois Urbana-Champaign), Huan Zhang (University of Illinois Urbana-Champaign)

CodeOptimizationComputational EfficiencyTransformerTabularBenchmark

🎯 What it does: This paper proposes the Clip-and-Verify framework, which utilizes linear constraints to trim the input domain and improve the intermediate layer boundaries during the Branch-and-Bound process of neural network verification, significantly reducing the number of subproblems and increasing the verification success rate.

CLIPGaussian: Universal and Multimodal Style Transfer Based on Gaussian Splatting

Kornel Howil (Jagiellonian University), PrzemysΕ‚aw Spurek (Jagiellonian University)

CodeImage TranslationGenerationOptimizationContrastive LearningGaussian SplattingImageVideoMultimodality

🎯 What it does: A general style transfer plugin based on Gaussian Splatting (CLIPGaussian) has been developed, which can achieve style transfer based on image or text prompts on 2D images, videos, 3D objects, and 4D dynamic scenes without increasing the model size.

Closed-Form Training Dynamics Reveal Learned Features and Linear Structure in Word2Vec-like Models

Dhruva Karkada (University of California Berkeley), Michael R DeWeese

CodeContrastive LearningText

🎯 What it does: This paper derives a quadratic word vector model called QWEM through a fourth-order Taylor approximation of the word2vec loss, and deduces its gradient flow training dynamics and the closed-form solution of the final embedding, revealing that the model learns orthogonal linear subspaces in a rank-increasing manner.

ClusterFusion: Expanding Operator Fusion Scope for LLM Inference via Cluster-Level Collective Primitive

Xinhao Luo (Shanghai Jiao Tong University), Minyi Guo (Shanghai Jiao Tong University)

CodeTransformerLarge Language ModelText

🎯 What it does: Developed the ClusterFusion framework, utilizing cluster-level communication primitives to achieve a broader range of operator fusion in LLM inference.

Co-Regularization Enhances Knowledge Transfer in High Dimensions

Shuo Shuo Liu (Pennsylvania State University), Runze Li (Pennsylvania State University)

CodeDomain AdaptationOptimizationSupervised Fine-TuningTabular

🎯 What it does: A common regularization transfer (CoRT) framework for multi-source high-dimensional generalized linear models is proposed, which directly imposes regularization constraints between the target and source models to achieve knowledge transfer.

Co-Reinforcement Learning for Unified Multimodal Understanding and Generation

Jingjing Jiang (Nanyang Technological University), Chao Ma (Shanghai Jiao Tong University)

CodeGenerationReinforcement LearningImageTextMultimodalityBenchmark

🎯 What it does: By proposing the CoRL framework, the understanding and generation capabilities of the unified multimodal model are enhanced through a joint reinforcement learning approach, ultimately resulting in ULM-R1.

COALA: Numerically Stable and Efficient Framework for Context-Aware Low-Rank Approximation

Uliana Parkina (Higher School of Economics University), Maxim Rakhuba (Higher School of Economics University)

CodeCompressionOptimizationTransformerLarge Language ModelSupervised Fine-TuningText

🎯 What it does: A context-aware low-rank approximation framework COALA based on regularization and without matrix inversion is proposed for compressing and fine-tuning large language models, maintaining numerical stability in scenarios with large calibration matrices and near-singular activation matrices.

CoCoA: A Minimum Bayes Risk Framework Bridging Confidence and Consistency for Uncertainty Quantification in LLMs

Roman Vashurin (Mohamed bin Zayed University of Artificial Intelligence), Maxim Panov (Mohamed bin Zayed University of Artificial Intelligence)

CodeTransformerLarge Language ModelText

🎯 What it does: A unified CoCoA framework is proposed, combining the confidence of LLMs with the semantic consistency of diverse outputs for uncertainty quantification;

Code Graph Model (CGM): A Graph-Integrated Large Language Model for Repository-Level Software Engineering Tasks

Hongyuan Tao (Ant Group), Peng Di (Ant Group)

CodeAI Code AssistantGraph Neural NetworkTransformerLarge Language ModelTextGraphBenchmarkRetrieval-Augmented Generation

🎯 What it does: Proposes the Code Graph Model (CGM) and the Graph RAG framework to achieve agent-free warehouse-level software engineering tasks;

CodeCrash: Exposing LLM Fragility to Misleading Natural Language in Code Reasoning

Man Ho LAM, Michael Lyu

CodeExplainability and InterpretabilityAI Code AssistantTransformerLarge Language ModelTextChain-of-Thought

🎯 What it does: Constructed the CODECRASH framework to stress test the robustness of LLM code reasoning through structural and natural language embedding perturbations.

CodeMerge: Codebook-Guided Model Merging for Robust Test-Time Adaptation in Autonomous Driving

Huitong Yang (University of Queensland), Yadan Luo (University of Queensland)

CodeDomain AdaptationAutonomous DrivingOptimizationPoint Cloud

🎯 What it does: CodeMerge is proposed, a model fusion framework based on low-dimensional fingerprint codebooks for adaptive 3D perception during online testing, addressing the unstable optimization problem in high-variance tasks.

Cognitive Mirrors: Exploring the Diverse Functional Roles of Attention Heads in LLM Reasoning

Xueqi Ma (University of Melbourne), James Bailey (University of Melbourne)

CodeTransformerLarge Language ModelText

🎯 What it does: This paper proposes an explanatory framework that identifies attention heads responsible for different cognitive functions (such as retrieval, logical reasoning, mathematical computation, etc.) in large language models through multi-class detection methods, and validates the key role of these 'cognitive heads' in reasoning tasks.

CoIDO: Efficient Data Selection for Visual Instruction Tuning via Coupled Importance-Diversity Optimization

Yichen Yan (Zhejiang University), Huan Li (Zhejiang University)

CodeOptimizationData-Centric LearningLarge Language ModelSupervised Fine-TuningImageMultimodality

🎯 What it does: Proposes the COIDO framework, which efficiently selects visual instruction fine-tuning data using the importance and diversity of joint optimization;

COLA: Towards Efficient Multi-Objective Reinforcement Learning with Conflict Objective Regularization in Latent Space

Pengyi Li (Tianjin University), YAN ZHENG

CodeOptimizationReinforcement Learning

🎯 What it does: This paper proposes the COLA framework for multi-objective reinforcement learning, utilizing shared latent space and conflict objective regularization to enhance sample efficiency and performance.

Collaborative Reasoner: Self-Improving Social Agents with Synthetic Conversations

Ansong Ni (Meta), Asli Celikyilmaz (Meta)

CodeTransformerLarge Language ModelReinforcement LearningTextChain-of-Thought

🎯 What it does: This paper proposes a collaborative reasoning framework called Coral, which utilizes multi-turn dialogue to evaluate and enhance the collaborative reasoning capabilities of LLMs;

Collapsing Taylor Mode Automatic Differentiation

Felix Dangel (Vector Institute), Andrea Walther (Humboldt Universitaet zu Berlin)

CodeOptimizationComputational EfficiencyAuto Encoder

🎯 What it does: This paper proposes a 'contraction Taylor mode' optimization technique for the computation of high-order derivatives (such as PDE operators), which combines linear structures to achieve graph rewriting, significantly reducing the required vector passing. It is applicable to linear PDEs such as Laplace, weighted Laplace, biharmonic, and their stochastic approximations.

Collective Counterfactual Explanations: Balancing Individual Goals and Collective Dynamics

Ahmad Reza Ehyaei, Samira Samadi (Max Planck Institute for Intelligent Systems)

CodeRecommendation SystemOptimizationExplainability and InterpretabilityTabular

🎯 What it does: This paper proposes a collective counterfactual explanation framework (CCE) that considers population dynamics and resource competition, capable of generating recommendations that minimize individual costs while also balancing group interests.

Color Conditional Generation with Sliced Wasserstein Guidance

Alexander Lobashev (Glam AI), Dmitry Guskov (Glam AI)

CodeImage TranslationGenerationData SynthesisDiffusion modelImageText

🎯 What it does: A training-free color conditional generation method named SW-Guidance is proposed, which uses the Sliced Wasserstein distance to guide the diffusion model to match the color distribution of reference images.

CoLT: The conditional localization test for assessing the accuracy of neural posterior estimates

Tianyu Chen (University of Texas at Austin), James G. Scott (University of Texas at Austin)

Code

🎯 What it does: A conditional location test (CoLT) is proposed to verify whether the posterior distribution output by the neural posterior estimator (NPE) is consistent with the true posterior.

COME: Adding Scene-Centric Forecasting Control to Occupancy World Model

Yining Shi (Tsinghua University), Diange Yang (Tsinghua University)

CodeGenerationAutonomous DrivingOptimizationTransformerDiffusion modelWorld ModelPoint Cloud

🎯 What it does: A framework named COME is proposed, which enhances the generation effect of occupancy world models through scene-centered predictive control.

ComfyMind: Toward General-Purpose Generation via Tree-Based Planning and Reactive Feedback

Litao Guo (Hong Kong University of Science and Technology), Ying-Cong Chen (Hong Kong University of Science and Technology)

CodeGenerationLarge Language ModelVision Language ModelImageVideoBenchmarkRetrieval-Augmented Generation

🎯 What it does: A collaborative AI system called ComfyMind has been built on the open-source ComfyUI platform to achieve general visual generation tasks (image and video generation and editing).

Communication-Efficient Diffusion Denoising Parallelization via Reuse-then-Predict Mechanism

Kunyun Wang (Shanghai Jiao Tong University), Jieru Zhao (Shanghai Jiao Tong University)

CodeRestorationComputational EfficiencyDiffusion modelImageAudio

🎯 What it does: Two methods, ParaStep and BatchStep, are proposed to accelerate diffusion model inference through a reuse-repredict mechanism, significantly reducing communication overhead.

Compact Memory for Continual Logistic Regression

Yohan Jung (RIKEN Center for AI Project), Mohammad Emtiyaz Khan (Imperial College London)

CodeClassificationOptimizationImage

🎯 What it does: A PPCA method based on Hessian matching is proposed to construct compact memory (memory vectors and weights) for logistic regression in continual learning, and to achieve approximate reconstruction of past task gradients through this memory, thereby alleviating catastrophic forgetting.

Competitive Advantage Attacks to Decentralized Federated Learning

Yuqi Jia (Duke University), Neil Zhenqiang Gong

CodeOptimizationFederated LearningAdversarial AttackImageText

🎯 What it does: This paper proposes SelfishAttack, targeting selfish clients in decentralized federated learning (DFL). It customizes the shared model to enhance the accuracy of its own model while suppressing the competitive advantage of non-selfish clients.

Compiler-R1: Towards Agentic Compiler Auto-tuning with Reinforcement Learning

Haolin Pan (Hangzhou Institute for Advanced Study), Yanjun Wu (Institute of Software Chinese Academy of Sciences)

CodeOptimizationTransformerLarge Language ModelSupervised Fine-TuningReinforcement LearningChain-of-Thought

🎯 What it does: Developed the Compiler-R1 framework, which utilizes reinforcement learning to drive large language models (LLM) for automatic compiler tuning, automatically generating optimization pass sequences for LLVM IR.

Complexity Scaling Laws for Neural Models using Combinatorial Optimization

Lowell Weissman (Virginia Tech), A. Lynn Abbott (Virginia Tech)

CodeOptimizationTransformerSupervised Fine-TuningReinforcement LearningTabularBenchmark

🎯 What it does: This paper conducts large-scale experiments on the Traveling Salesman Problem (TSP) to study the scalability of neural networks under fixed model capacity, proposing a 'problem complexity scaling law' based on the size of the solution space and the representation space.

Composing Global Solutions to Reasoning Tasks via Algebraic Objects in Neural Nets

Yuandong Tian (Meta Superintelligence Lab)

CodeOptimizationTabular

🎯 What it does: This paper studies a two-dimensional hidden layer neural network trained on Abelian group reasoning tasks (such as modular addition), revealing the semiring algebraic structure of its weight space and loss function, and constructs a global optimal solution based on this;

Composing Linear Layers from Irreducibles

Travis Pence (University of Wisconsin-Madison), Vikas Singh (University of Wisconsin-Madison)

CodeCompressionComputational EfficiencyTransformerLarge Language ModelText

🎯 What it does: Using Clifford algebra to decompose linear layers into geometric primitives (bivectors) and achieve differentiable linear mappings through rotor sandwiching, forming a parameter-compressed linear module.

Composite Flow Matching for Reinforcement Learning with Shifted-Dynamics Data

Lingkai Kong (Harvard University), Milind Tambe (Harvard University)

CodeReinforcement LearningFlow-based ModelTabular

🎯 What it does: To address the dynamic shift between offline and online data in reinforcement learning, the COMPFLOW method is proposed, which uses combined flow matching to estimate the dynamic gap and designs an active exploration strategy based on the gap results.

Composition and Alignment of Diffusion Models using Constrained Learning

Shervin Khalafi (University of Pennsylvania), Alejandro Ribeiro (University of Pennsylvania)

CodeGenerationOptimizationDiffusion modelImageText

🎯 What it does: This paper proposes a unified constrained learning framework for aligning and composing diffusion models, generating samples that meet multiple constraints by imposing constraints on the KL divergence between pre-trained models and reward functions or multiple pre-trained models.

Compositional Discrete Latent Code for High Fidelity, Productive Diffusion Models

Samuel Lavoie (Mila UniversitΓ© de MontrΓ©al), Aaron Courville (Mila UniversitΓ© de MontrΓ©al)

CodeGenerationData SynthesisTransformerLarge Language ModelDiffusion modelImageText

🎯 What it does: This paper proposes and utilizes Discrete Latent Code (DLC) to condition diffusion models, achieving high-fidelity unconditional image generation on ImageNet, and enabling image composability and text-to-image generation through DLC.

Compositional Neural Network Verification via Assume-Guarantee Reasoning

Hai Duong (George Mason University), Matthew B. Dwyer (University of Virginia)

CodeComputational EfficiencyConvolutional Neural NetworkAuto EncoderImage

🎯 What it does: A parameterizable hypothesis-guarantee compositional framework called CoVeNN is proposed to decompose large-scale neural networks into sub-networks and verify them incrementally, significantly reducing memory consumption and enhancing verifiability.

Computational Algebra with Attention: Transformer Oracles for Border Basis Algorithms

Hiroshi Kera (Chiba University), Sebastian Pokutta (Zuse Institute Berlin)

CodeComputational EfficiencyTransformer

🎯 What it does: A new boundary basis algorithm is proposed, utilizing deep learning to accelerate the solving of polynomial equation systems while ensuring the correctness of the output.

ComRank: Ranking Loss for Multi-Label Complementary Label Learning

Jing-Yi Zhu (Southeast University), Min-Ling Zhang (University of Queensland)

CodeClassificationOptimizationImageTextBiomedical Data

🎯 What it does: A ranking loss framework called ComRank is proposed for multi-label complementary label learning, addressing the failure of traditional unbiased risk estimation methods under non-uniform complementary label distributions.

Concept Incongruence: An Exploration of Time and Death in Role Playing

Xiaoyan Bai (University of Chicago), Chenhao Tan (University of Chicago)

CodeTransformerLarge Language ModelPrompt EngineeringText

🎯 What it does: This paper studies the conceptual inconsistency caused by the conflict between the time of character death and query time in role-playing scenarios using large language models. It proposes three behavioral metrics (abandonment rate, response rate, conditional accuracy) and reveals the model's insufficient internal encoding of death states and times, leading to a decrease in accuracy through dialogue experiments and linear probe analysis.

ConceptScope: Characterizing Dataset Bias via Disentangled Visual Concepts

Jinho Choi (KAIST), Jaegul Choo (KAIST)

CodeClassificationData-Centric LearningLarge Language ModelVision Language ModelAuto EncoderImageMultimodality

🎯 What it does: This paper proposes a framework called ConceptScope for the automatic discovery and quantification of visual concepts in image datasets, aimed at identifying dataset biases and assessing model robustness.

Conditional Diffusion Anomaly Modeling on Graphs

Chunyu Wei (Renmin University of China), Yunhai Wang (Renmin University of China)

CodeAnomaly DetectionGraph Neural NetworkDiffusion modelGraphFinance Related

🎯 What it does: A graph anomaly detection method based on a conditional diffusion model, CGADM, is proposed, transforming anomaly detection into generative modeling.

Conditional Gradient Methods with Standard LMO for Stochastic Simple Bilevel Optimization

Khanh-Hung Giang-Tran (Cornell University), Nam Ho-Nguyen (University of Sydney)

CodeOptimizationHyperparameter SearchTabular

🎯 What it does: An iterative regularization projection-free conditional gradient algorithm is proposed to solve a single-layer simple two-layer optimization problem with randomness using a linear optimization oracle.

Conformal Information Pursuit for Interactively Guiding Large Language Models

Kwan Ho Ryan Chan (University of Pennsylvania), Rene Vidal

CodeTransformerLarge Language ModelSupervised Fine-TuningPrompt EngineeringText

🎯 What it does: This paper proposes an interactive question-answering framework called Conformal Information Pursuit (C-IP), which measures uncertainty by using the size of the prediction set in large language models (LLMs) to select the most informative queries at each step, thereby reducing the number of queries and improving prediction accuracy.

Conformal Online Learning of Deep Koopman Linear Embeddings

Ben Gao (UniversitΓ© Jean Monnet Saint-Etienne), Olivier Alata (UniversitΓ© Jean Monnet Saint-Etienne)

CodeTabularTime Series

🎯 What it does: An online learning framework named COLoKe is proposed, which adaptively updates deep Koopman embeddings using a conformal mechanism.

Conformal Prediction Beyond the Seen: A Missing Mass Perspective for Uncertainty Quantification in Generative Models

Sima Noorani (University of Pennsylvania), Hamed Hassani (University of Pennsylvania)

CodeGenerationData SynthesisTransformerLarge Language ModelText

🎯 What it does: A conformal prediction framework CPQ is proposed from the perspective of missing mass to quantify uncertainty in situations where only generative models can be queried.

Conformal Prediction for Ensembles: Improving Efficiency via Score-Based Aggregation

Yash Patel, Ambuj Tewari

CodeOptimizationComputational EfficiencyScore-based ModelTabular

🎯 What it does: A conformal score aggregation (CSA) framework based on multidimensional scores and quantile envelopes is proposed to generate efficient prediction intervals without distributional assumptions in ensemble models.

Conformal Prediction for Time-series Forecasting with Change Points

Sophia Huiwen Sun (University of California San Diego), Rose Yu (University of California San Diego)

CodeAnomaly DetectionOptimizationTime Series

🎯 What it does: The CPTC algorithm is proposed, which combines state prediction from state space models with online shape prediction to generate reliable prediction intervals on time series with change points.

Conformal Prediction in The Loop: A Feedback-Based Uncertainty Model for Trajectory Optimization

Han Wang (Shanghai Jiao Tong University), Chao Ning (Shanghai Jiao Tong University)

CodeAutonomous DrivingOptimizationRecurrent Neural NetworkReinforcement LearningTime SeriesSequential

🎯 What it does: A feedback-based conformal prediction (Fb-CP) framework is proposed for trajectory optimization in uncertain environments, capable of real-time adjustment of the prediction region and risk allocation.

Conformal Risk Training: End-to-End Optimization of Conformal Risk Control

Christopher Yeh (California Institute of Technology), Yisong Yue (California Institute of Technology)

CodeSegmentationOptimizationImageBiomedical Data

🎯 What it does: A distribution-independent, finite-sample risk control mechanism for optimizing coherent risk equivalence (OCE) (including CVaR) is proposed, which is embedded into the model training process, forming a 'synthetic risk training' method.

ConfTuner: Training Large Language Models to Express Their Confidence Verbally

Yibo Li (National University of Singapore), Bryan Hooi (National University of Singapore)

CodeTransformerLarge Language ModelSupervised Fine-TuningText

🎯 What it does: The ConfTuner framework is proposed, which fine-tunes the tokenized Brier loss on LLMs, enabling the model to express confidence that aligns with actual reliability.

Confusion-Driven Self-Supervised Progressively Weighted Ensemble Learning for Non-Exemplar Class Incremental Learning

Kai Hu (Xiangtan University), Xieping Gao (Hunan Normal University)

CodeRepresentation LearningConvolutional Neural NetworkContrastive LearningImage

🎯 What it does: The CLOVER framework is proposed to address the issue of representation overlap between old and new classes in non-example class incremental learning (NECIL);

Connecting Jensen–Shannon and Kullback–Leibler Divergences: A New Bound for Representation Learning

Reuben Dorent, William M Wells

CodeOptimizationRepresentation LearningTabularSequential

🎯 What it does: A new lower bound of KLD for JSD is proposed, proving that maximizing JSD is equivalent to optimizing the lower bound of mutual information, and the relationship between cross-entropy loss and JSD is provided.

Connecting Neural Models Latent Geometries with Relative Geodesic Representations

Hanlin Yu, Marco Fumero (Institute of Science and Technology Austria)

CodeGenerationRetrievalRepresentation LearningAuto EncoderImageMultimodality

🎯 What it does: This study investigates the parameterization of different neural networks learning the same latent manifold on similar data and proposes a method to construct relative geodesic representations using pulled-back Riemannian metrics to achieve unsupervised communication between latent spaces.

Consensus-Robust Transfer Attacks via Parameter and Representation Perturbations

Shixin Li (Huazhong University of Science and Technology), Bin Benjamin Zhu (Microsoft Corporation)

CodeAdversarial AttackConvolutional Neural NetworkTransformerImage

🎯 What it does: A transfer attack based on consistent robustness, CORTA, is proposed, which enhances the transferability of black-box attacks using parameter perturbation and representation mixing.

Consistent Paths Lead to Truth: Self-Rewarding Reinforcement Learning for LLM Reasoning

Kongcheng Zhang (Zhejiang University), Dacheng Tao (Nanyang Technological University)

CodeTransformerLarge Language ModelReinforcement LearningText

🎯 What it does: A self-rewarding reinforcement learning framework called CoVo is proposed, which constructs internal rewards by utilizing the consistency and volatility of intermediate states in LLM reasoning trajectories, enhancing reasoning capabilities without external supervision.

Consistent Sampling and Simulation: Molecular Dynamics with Energy-Based Diffusion Models

Michael Plainer (Freie UniversitΓ€t Berlin), Frank Noe (Microsoft Research)

CodeOptimizationDrug DiscoveryGraph Neural NetworkTransformerMixture of ExpertsDiffusion modelScore-based ModelBiomedical Data

🎯 What it does: This study trained an energy-based diffusion model and achieved consistency between sampling and simulation by incorporating Fokker-Planck regularization. It also proposed a Mixture of Experts method for time periods and constructed a transferable dipeptide Boltzmann simulator, applying this method to coarse-grained biomolecules such as dipeptides, Chignolin, and BBA.

Constrained Entropic Unlearning: A Primal-Dual Framework for Large Language Models

Taha Entesari (Johns Hopkins University), Mahyar Fazlyab (Johns Hopkins University)

CodeOptimizationTransformerLarge Language ModelText

🎯 What it does: This paper proposes a framework that views the 'forgetting' problem of large language models as a constrained optimization problem, and achieves uniformity of the forgetting set through a new 'logit-margin flattening' loss; it employs a scalable primal-dual algorithm to achieve efficient forgetting while maintaining the performance of the retained set.

Constrained Sampling for Language Models Should Be Easy: An MCMC Perspective

Emmanuel Anaya Gonzalez (University of California San Diego), Loris D'Antoni (University of California San Diego)

CodeGenerationOptimizationLarge Language ModelReinforcement LearningText

🎯 What it does: A constraint sampling framework based on MCMC is proposed, allowing language models to efficiently and feasibly sample from constrained distributions while satisfying context-free grammar constraints.

Context-Aware Hierarchical Learning: A Two-Step Paradigm towards Safer LLMs

Tengyun Ma (Harbin Institute of Technology), Zhuotao Tian (Harbin Institute of Technology)

CodeSafty and PrivacyAdversarial AttackTransformerLarge Language ModelSupervised Fine-TuningPrompt EngineeringTextBenchmark

🎯 What it does: This paper studies prompt injection attacks in the context of tool invocation, proposing a novel Tool-Completion Attack (TCA), constructing a corresponding benchmark dataset, and designing a Context-Aware Hierarchical Learning (CAHL) mechanism to enhance the security of large language models.

Context-Aware Regularization with Markovian Integration for Attention-Based Nucleotide Analysis

Mohammadsaleh Refahi, Gail Rosen (Drexel University)

CodeTransformerSupervised Fine-TuningBiomedical Data

🎯 What it does: Designed and trained the CARMANIA model, achieving genome sequence modeling of up to 160kbp using sliding window attention and transfer matrix (TM) loss;

ContextAgent: Context-Aware Proactive LLM Agents with Open-world Sensory Perceptions

Bufang Yang (Chinese University of Hong Kong), Zhenyu Yan (Chinese University of Hong Kong)

CodeTransformerLarge Language ModelSupervised Fine-TuningVision Language ModelVideoTextMultimodalityBenchmarkChain-of-ThoughtAudio

🎯 What it does: This paper proposes ContextAgent, a LLM agent that utilizes multimodal perception from wearable devices for context awareness and actively invokes tools, along with the design of a corresponding evaluation benchmark, ContextAgentBench.

ConTextTab: A Semantics-Aware Tabular In-Context Learner

Marco Spinaci (SAP France), Sam Thelin (SAP SE)

CodeTransformerTabular

🎯 What it does: A semantic-aware In-Context learning model called ConTextTab is proposed, which is based on a table-native architecture. It utilizes real-world table data for pre-training and employs multi-modal embeddings such as column names, text, and dates during prediction, balancing scalability and semantic understanding.

Continual Gaussian Mixture Distribution Modeling for Class Incremental Semantic Segmentation

Guilin Zhu (Huazhong University of Science and Technology), Changxin Gao (Huazhong University of Science and Technology)

CodeSegmentationKnowledge DistillationConvolutional Neural NetworkImage

🎯 What it does: The CoGaMiD method is proposed, which utilizes Gaussian Mixture Model to perform multimodal modeling of the feature distribution of learned classes and dynamically updates it during the incremental learning process to achieve class-incremental semantic segmentation.

Continual Knowledge Adaptation for Reinforcement Learning

Jinwu Hu (South China University of Technology), Mingkui Tan (South China University of Technology)

CodeReinforcement Learning

🎯 What it does: The CKA-RL method is proposed, which maintains a task-specific knowledge vector pool and dynamically utilizes historical knowledge for adaptation, achieving knowledge accumulation and transfer in continuous reinforcement learning while reducing catastrophic forgetting.

Continual Multimodal Contrastive Learning

Xiaohao Liu (National University of Singapore), Tat-Seng Chua (National University of Singapore)

CodeClassificationRetrievalRepresentation LearningContrastive LearningVideoMultimodalityAudio

🎯 What it does: This paper proposes a formal framework for Continuous Multimodal Contrastive Learning (CMCL), defining two main objectives: stability and plasticity. It achieves incremental learning of new modality data at the gradient level based on the Dual-Sided Null Space (DNS) method while retaining knowledge of old modalities.

Continual Optimization with Symmetry Teleportation for Multi-Task Learning

Zhipeng Zhou (Nanyang Technological University), Chunyan Miao (Nanyang Technological University)

CodeOptimizationSupervised Fine-TuningImage

🎯 What it does: The research proposes a continuous optimization framework COST based on symmetric moment transfer and LoRA to address gradient conflicts and task imbalance in multi-task learning.

Continuous Diffusion Model for Language Modeling

Jaehyeong Jo (Korea Advanced Institute of Science and Technology), Sung Ju Hwang (DeepAuto.ai)

CodeGenerationData SynthesisTransformerDiffusion modelImageTextBiomedical Data

🎯 What it does: A continuous diffusion model RDLM based on Riemannian statistical manifolds is proposed, which can map discrete tokens onto the orthogonal sphere and design a continuous diffusion process on that sphere to achieve parallel generation of language, images, and biological sequences.

Continuous Domain Generalization

Zekun Cai (Jilin University), Liang Zhao (Emory University)

CodeDomain AdaptationAdversarial AttackMeta LearningNeural Radiance FieldTabularSequential

🎯 What it does: This paper proposes the Continuous Domain Generalization (CDG) task and achieves continuous transfer of model parameters in a multi-dimensional continuous domain space through the Neural Lie transport operator, addressing the issue of continuous evolution between domains.

Continuous Simplicial Neural Networks

Aref Einizade (Telecom Paris Institute Polytechnique de Paris), Jhony H. Giraldo (Telecom Paris Institute Polytechnique de Paris)

CodeGraph Neural NetworkGraphOrdinary Differential Equation

🎯 What it does: A continuous complex-valued neural network (COSIMO) is proposed, utilizing partial differential equations to achieve information propagation on complex numbers;

Continuous Soft Actor-Critic: An Off-Policy Learning Method Robust to Time Discretization

Huimin Han (Zhongtai Securities Institute for Financial Studies Shandong University), Shaolin Ji (Zhongtai Securities Institute for Financial Studies Shandong University)

CodeReinforcement LearningStochastic Differential Equation

🎯 What it does: This paper proposes the Continuous Soft Actor-Critic (CSAC) and its multi-agent extension (CMASAC) for offline reinforcement learning in continuous-time stochastic environments, addressing the sensitivity issue of time discretization.

Contrastive Self-Supervised Learning As Neural Manifold Packing

Guanming Zhang (New York University), Stefano Martiniani (New York University)

CodeClassificationObject DetectionRepresentation LearningConvolutional Neural NetworkContrastive LearningImage

🎯 What it does: This paper proposes the CLAMP framework, which reinterprets contrastive self-supervised learning as a neural manifold packing problem, and constructs a unidirectional loss function through short-range repulsive potential, dynamically optimizing the size and position of the enhancement sub-manifold for each image.

ControlFusion: A Controllable Image Fusion Network with Language-Vision Degradation Prompts

Linfeng Tang (Wuhan University), Jiayi Ma (Macau University of Science and Technology)

CodeRestorationObject DetectionTransformerPrompt EngineeringImageMultimodality

🎯 What it does: A controllable image fusion network called ControlFusion is proposed, which utilizes language-visual prompts to achieve adaptive denoising and fusion for composite degradation.

Controlling Thinking Speed in Reasoning Models

Zhengkai Lin (Zhejiang University), Jieping Ye (Alibaba Cloud)

CodeLarge Language ModelPrompt EngineeringText

🎯 What it does: A method is proposed for dynamically controlling the thinking speed of large reasoning models (LRM) during inference, utilizing the transition vectors of the model's internal representations to switch between different thinking modes (fast intuitive and slow deep) and dynamically adjusting the reasoning speed through real-time difficulty estimation.

Convergence Rates of Constrained Expected Improvement

Haowei Wang (National University of Singapore), Cosmin G. Petra (Lawrence Livermore National Laboratory)

CodeOptimizationTabular

🎯 What it does: This paper studies the convergence rate of the Constrained Expected Improvement (CEI) algorithm, establishing for the first time an upper bound on its simple regret and analyzing its convergence under both frequentist and Bayesian assumptions.

Cooperative Retrieval-Augmented Generation for Question Answering: Mutual Information Exchange and Ranking by Contrasting Layers

Youmin Ko (Hanyang University), Hyunjoon Kim (Hanyang University)

CodeGenerationRetrievalTransformerLarge Language ModelTextRetrieval-Augmented Generation

🎯 What it does: Proposes the CoopRAG framework, which first breaks down the problem into sub-problems and uncertain reasoning chains, utilizes a retriever to search for documents, performs hierarchical comparative re-ranking of the search results, and finally allows the LLM to fill in missing information and provide answers.

CORAL: Disentangling Latent Representations in Long-Tailed Diffusion

Esther Rodriguez (Arizona State University), Lalitha Sankar (Arizona State University)

CodeGenerationData SynthesisDiffusion modelContrastive LearningImage

🎯 What it does: This paper proposes an improved diffusion model method called CORAL for long-tail data distribution, addressing the issue of low-quality generation in tail classes.

CORE: Collaborative Optimization with Reinforcement Learning and Evolutionary Algorithm for Floorplanning

Pengyi Li (Tianjin University), Junchi Yan (Shanghai Jiao Tong University)

CodeOptimizationReinforcement Learning

🎯 What it does: A hybrid evolutionary reinforcement learning framework called CORE based on B*-Tree representation is proposed to solve the floorplanning problem in electronic design automation.

CORE: Reducing UI Exposure in Mobile Agents via Collaboration Between Cloud and Local LLMs

Gucongcong Fan (Shanghai Jiao Tong University), Guihai Chen (Shanghai Jiao Tong University)

CodeTransformerLarge Language ModelTextBenchmark

🎯 What it does: The CORE framework is proposed, which enables collaboration between a local lightweight LLM and a powerful cloud LLM to minimize the upload of mobile UI information while maintaining task success rates.

Corporate Needs You to Find the Difference: Revisiting Submodular and Supermodular Ratio Optimization Problems

Elfarouk Harb (University of Illinois at Urbana Champaign), Chandra Chekuri (University of Illinois at Urbana Champaign)

CodeRecommendation SystemOptimizationComputational EfficiencyGraphFinance Related

🎯 What it does: This paper proposes a unified 'universal solver'β€”SUPERGREEDY++, Frank-Wolfe, and Fujishige-Wolfe FW-MNPβ€”by proving the exact and approximate equivalence of SFM, DSS, USSS, UDSS with the minimum norm point (MNP), which can directly solve multi-class sub/supermodular ratio problems and conduct large-scale experiments on tasks such as HNSN and minimum s-t-cut.

Correcting misinterpretations of additive models

Benedict Clark (Physikalisch-Technische Bundesanstalt), Stefan Haufe (CharitΓ© - UniversitΓ€tsmedizin)

CodeExplainability and InterpretabilityImageTabularElectronic Health Records

🎯 What it does: Two activation pattern-based explanation methods, PatternGAM and PatternQLR, are proposed to correct misinterpretations in additive models, particularly to suppress errors in variable importance caused by confounding variables.

Correlated Low-Rank Adaptation for ConvNets

Wu Ran (Shanghai Jiao Tong University), Chao Ma (Shanghai Jiao Tong University)

CodeDomain AdaptationOptimizationConvolutional Neural NetworkSupervised Fine-TuningImage

🎯 What it does: A parameter-efficient fine-tuning method for convolutional networks called CoLoRA is proposed, which utilizes low-rank shared matrices of adjacent layers to achieve correlation learning.

Cost-Sensitive Freeze-thaw Bayesian Optimization for Efficient Hyperparameter Tuning

Dong Bok Lee (Korea Advanced Institute of Science and Technology), Hae Beom Lee (Korea University)

CodeOptimizationHyperparameter SearchTransformerTabular

🎯 What it does: A cost-sensitive freeze-thaw Bayesian optimization method is proposed to achieve early stopping in multi-precision hyperparameter optimization based on a user-defined utility function.

CoT Red-Handed: Stress Testing Chain-of-Thought Monitoring

Benjamin Arnav (LASR Labs), Mary Phuong (LASR Labs)

CodeAnomaly DetectionAI Code AssistantLarge Language ModelTextChain-of-Thought

🎯 What it does: This paper studies the effectiveness of chain-of-thought (CoT) monitoring compared to action-only review in detecting potential malicious behavior in AI control frameworks, and proposes a hybrid monitoring scheme that integrates both approaches.

Counterfactual Implicit Feedback Modeling

Chuan Zhou (University of Melbourne), Mingming Gong (University of Melbourne)

CodeRecommendation SystemContrastive LearningTabular

🎯 What it does: Modeling the relevance prediction problem of implicit feedback as a counterfactual estimation problem with missing data handling, and proposing the Counter-IF method.

Counterfactual Reasoning for Steerable Pluralistic Value Alignment of Large Language Models

Hanze Guo (Renmin University of China), Xing Xie (Microsoft Research)

CodeTransformerLarge Language ModelSupervised Fine-TuningPrompt EngineeringText

🎯 What it does: Proposes the COUPLE framework, which utilizes structural causal models and counterfactual reasoning to achieve multi-dimensional value alignment for LLMs, supporting adjustable fine-grained value priorities;

Counterfactual reasoning: an analysis of in-context emergence

Moritz Miller (Max Planck Institute for Intelligent Systems), Siyuan Guo (University of Cambridge)

CodeTransformerLarge Language ModelTime SeriesSequentialStochastic Differential Equation

🎯 What it does: This paper studies the ability of large-scale language models (especially Transformers) to perform counterfactual reasoning in context, designing synthetic linear regression tasks and SDE-based dynamical system tasks to evaluate how models achieve noise elimination and counterfactual prediction through self-attention and depth, providing corresponding theoretical and mechanistic explanations.

Coupled Data and Measurement Space Dynamics for Enhanced Diffusion Posterior Sampling

Shayan Mohajer Hamidi (Stanford University), EN-HUI YANG

CodeRestorationDiffusion modelImageBenchmark

🎯 What it does: A new framework called Coupled Data and Measurement Space Diffusion Posterior Sampling (C-DPS) is proposed to address signal recovery in inverse problems, particularly in the presence of noise or incomplete measurements.

Covariate-moderated Empirical Bayes Matrix Factorization

William R.P. Denault, Matthew Stephens (University of Chicago)

CodeRecommendation SystemTabular

🎯 What it does: A new covariate-modulated empirical Bayesian matrix factorization (cEBMF) framework is proposed, which can utilize any form of side information to improve matrix factorization.

Covering Multiple Objectives with a Small Set of Solutions Using Bayesian Optimization

Natalie Maus (University of Pennsylvania), Jacob R. Gardner (University of Pennsylvania)

CodeOptimizationDrug DiscoveryTabular

🎯 What it does: A Bayesian optimization framework for black-box multi-objective coverage optimization, MOCOBO, is proposed, which can find K solutions that cover all objectives in high-dimensional structured search spaces.

CovMatch: Cross-Covariance Guided Multimodal Dataset Distillation with Trainable Text Encoder

Yongmin Lee (KAIST), Hye Won Chung (KAIST)

CodeRetrievalKnowledge DistillationVision Language ModelContrastive LearningImageTextMultimodality

🎯 What it does: This study investigates how to perform multimodal dataset distillation in large-scale vision-language models to generate a small number of high-quality image-text pairs.

CoVoMix2: Advancing Zero-Shot Dialogue Generation with Fully Non-Autoregressive Flow Matching

leying zhang, sheng zhao

CodeGenerationTransformerFlow-based ModelAudio

🎯 What it does: Developed CoVoMix2, a fully non-autoregressive zero-shot multi-speaker dialogue generation framework that directly predicts mixed Mel spectrograms from multi-stream text, supporting overlapping speech and temporal control.

CPPO: Accelerating the Training of Group Relative Policy Optimization-Based Reasoning Models

ZhiHang Lin, Rongrong Ji (Xiamen University)

CodeOptimizationComputational EfficiencyReinforcement LearningAgentic AITabular

🎯 What it does: This study investigates the acceleration of training inference models based on GRPO through pruning high-advantage completions, proposing CPPO and a dynamic completion allocation strategy.

CQ-DINO: Mitigating Gradient Dilution via Category Queries for Vast Vocabulary Object Detection

Zhichao Sun (Wuhan University), Yongchao Xu (Wuhan University)

CodeObject DetectionTransformerContrastive LearningImage

🎯 What it does: The CQ-DINO framework is proposed to address the issue of positive and negative gradient dilution in large vocabulary object detection through learnable category queries and image-guided query selection.

Creativity or Brute Force? Using Brainteasers as a Window into the Problem-Solving Abilities of Large Language Models

Simeng Han (Yale University), R. Thomas McCoy (Yale University)

CodeTransformerLarge Language ModelPrompt EngineeringTextBenchmarkChain-of-Thought

🎯 What it does: Designed and evaluated the BRAINGLE BRAINTEASER benchmark, which systematically measures the reasoning process, creativity, tendency for brute-force search, step decomposition, and prompt utilization of LLMs in long narrative brainteasers.

Credal Prediction based on Relative Likelihood

Timo LΓΆhr (LMU Munich), Eyke HΓΌllermeier (LMU Munich)

CodeAnomaly DetectionNeural Architecture SearchConvolutional Neural NetworkImageMagnetic Resonance Imaging

🎯 What it does: A trustworthy prediction method based on relative likelihood thresholds is proposed, and an approximately adjustable trustworthy set is integrated through neural networks.

CReFT-CAD: Boosting Orthographic Projection Reasoning for CAD via Reinforcement Fine-Tuning

Ke Niu (Fudan University), Xiangyang Xue (Fudan University)

CodeTransformerReinforcement LearningVision Language ModelMultimodalityBenchmarkChain-of-Thought

🎯 What it does: A two-stage fine-tuning framework called CReFT-CAD is proposed, which combines curriculum-driven reinforcement learning with supervised post-tuning to enhance CAD projection reasoning capabilities, and a large-scale projection reasoning benchmark called TriView2CAD is released.

CroPe: Cross-Modal Semantic Compensation Adaptation for All Adverse Scene Understanding

Qin Xu (Anhui University), Bo Jiang (Anhui University)

CodeSegmentationDomain AdaptationAutonomous DrivingTransformerPrompt EngineeringVision Language ModelImageMultimodality

🎯 What it does: A cross-modal semantic compensation adaptation method called CroPe is proposed, which enhances the performance of semantic segmentation tasks in unsafe domains under extreme weather conditions using visual + text.

Cross City Traffic Flow Generation via Retrieval Augmented Diffusion Model

Yudong Li (Beihang University), Qian Huang (Huawei Technologies Co., Ltd)

CodeGenerationDomain AdaptationTransformerDiffusion modelTime SeriesRetrieval-Augmented Generation

🎯 What it does: Proposed a cross-city zero-shot traffic flow generation model CRAFT

CrossAD: Time Series Anomaly Detection with Cross-scale Associations and Cross-window Modeling

Beibu Li (East China Normal University), Chenjuan Guo (East China Normal University)

CodeAnomaly DetectionTransformerTime Series

🎯 What it does: Proposes the CrossAD framework, which utilizes cross-scale reconstruction and cross-window modeling for time series anomaly detection.

Crucible: Quantifying the Potential of Control Algorithms through LLM Agents

Lianchen Jia (Tsinghua University), Lifeng Sun (Tsinghua University)

CodeOptimizationReinforcement Learning from Human FeedbackTransformerLarge Language ModelReinforcement LearningTabularTime Series

🎯 What it does: The Crucible framework is proposed, which systematically evaluates and quantifies the tuning potential of control algorithms through LLM-driven multi-layer expert simulation and formal tunability metrics.

CryptoMoE: Privacy-Preserving and Scalable Mixture of Experts Inference via Balanced Expert Routing

Yifan Zhou (Peking University), Meng Li (Peking University)

CodeSafty and PrivacyComputational EfficiencyTransformerLarge Language ModelMixture of ExpertsText

🎯 What it does: This paper proposes the CryptoMoE framework, which implements private inference for sparse gated Mixture of Experts (MoE) large language models;