NeurIPS 2024 Papers — Page 7
Conference on Neural Information Processing Systems · 4035 papers
Clustering with Non-adaptive Subset Queries
Hadley Black (University of California San Diego), Barna Saha (University of California San Diego)
🎯 What it does: This paper proposes a novel non-adaptive subset query framework for accurately recovering any k-clustering without the need for pairwise queries.
CNCA: Toward Customizable and Natural Generation of Adversarial Camouflage for Vehicle Detectors
Linye Lyu (Harbin Institute of Technology), YU LI
Object DetectionAutonomous DrivingAdversarial AttackDiffusion modelImage
🎯 What it does: A customizable and natural vehicle detector adversarial camouflage method (CNCA) based on a pre-trained diffusion model is designed, which can generate diverse camouflage textures through user text prompts while maintaining high attack performance.
Co-occurrence is not Factual Association in Language Models
Xiao Zhang (Tsinghua University), Ji Wu (Tsinghua University)
TransformerLarge Language ModelSupervised Fine-TuningText
🎯 What it does: The study investigates that language models tend to learn word co-occurrence statistics rather than real factual associations during the fine-tuning process, and proposes improvements to enhance the generalizability of factual knowledge.
Coarse-to-Fine Concept Bottleneck Models
Konstantinos P. Panousis (University of Athens), Diego Marcos (Inria)
ClassificationExplainability and InterpretabilityConvolutional Neural NetworkVision Language ModelImage
🎯 What it does: A Coarse-Fine Concept Bottleneck Model (CF-CBM) is proposed, which enhances interpretability and classification performance in visual tasks by simultaneously conducting concept discovery on the entire image and image patches.
CoBo: Collaborative Learning via Bilevel Optimization
Diba Hashemi (École Polytechnique Fédérale de Lausanne), Martin Jaggi (École Polytechnique Fédérale de Lausanne)
OptimizationFederated LearningText
🎯 What it does: This paper models collaborative learning as a bi-level optimization problem and proposes the COBO algorithm to achieve alternating optimization of client selection and model training.
CODA: A Correlation-Oriented Disentanglement and Augmentation Modeling Scheme for Better Resisting Subpopulation Shifts
Ziquan OU (City University of Hong Kong), Zijun Zhang
Data SynthesisDomain AdaptationAuto EncoderImage
🎯 What it does: A CODA framework is proposed, utilizing correlation-driven decomposition and enhancement techniques to improve the model's robustness under subgroup distribution shifts (SC-GI).
Code Repair with LLMs gives an Exploration-Exploitation Tradeoff
Hao Tang (Cornell University), Kevin Ellis (Cornell University)
OptimizationAI Code AssistantTransformerLarge Language ModelReinforcement LearningText
🎯 What it does: This paper proposes an iterative code repair method called REx based on large language models, which models the repair process as an 'arm-acquiring' multi-armed bandit, intelligently balancing exploration and exploitation, significantly improving program generation efficiency.
CODE: Contrasting Self-generated Description to Combat Hallucination in Large Multi-modal Models
Junho Kim (Korea Advanced Institute of Science and Technology), Yong Man Ro (Korea Advanced Institute of Science and Technology)
GenerationData SynthesisTransformerLarge Language ModelVision Language ModelContrastive LearningTextMultimodalityBenchmark
🎯 What it does: A training-free contrastive decoding method called CODE is proposed, which uses the complete descriptions generated by LMM as visual references to dynamically compare logits and adjust the information flow during the decoding phase, thereby suppressing cross-modal hallucinations.
Coded Computing for Resilient Distributed Computing: A Learning-Theoretic Framework
Parsa Moradi (University of Minnesota), Mohammad Ali Maddah-Ali (University of Minnesota)
Convolutional Neural NetworkTransformerImage
🎯 What it does: A fault-tolerant distributed computing framework called LeTCC is designed based on learning theory, utilizing optimizable encoding/decoding functions to achieve robustness against slow nodes (stragglers) and noise.
CodeRosetta: Pushing the Boundaries of Unsupervised Code Translation for Parallel Programming
Ali TehraniJamsaz (Iowa State University), Ali Jannesari (Iowa State University)
AI Code AssistantTransformerLarge Language ModelSupervised Fine-TuningText
🎯 What it does: Proposed and implemented the CodeRosetta model for unsupervised bidirectional translation between general programming languages and their high-performance parallel extensions (e.g., C++↔CUDA, Fortran↔C++).
Coevolving with the Other You: Fine-Tuning LLM with Sequential Cooperative Multi-Agent Reinforcement Learning
Hao Ma (School of Artificial Intelligence, University of Chinese Academy of Sciences), Min Chen (Institute of Automation, Chinese Academy of Sciences)
OptimizationTransformerLarge Language ModelReinforcement LearningAgentic AITextSequential
🎯 What it does: A collaborative multi-agent reinforcement learning-based LLM fine-tuning framework called CORY is proposed, utilizing two agents, a pioneer and an observer, for knowledge transfer and role exchange to achieve co-evolution.
CoFie: Learning Compact Neural Surface Representations with Coordinate Fields
Hanwen Jiang (University of Texas at Austin), Qixing Huang (University of Texas at Austin)
CompressionRepresentation LearningPoint CloudMesh
🎯 What it does: This paper proposes CoFie, which utilizes a local coordinate framework and MLP for compressible neural implicit surface representation, significantly compressing parameters while maintaining detail.
CogVLM: Visual Expert for Pretrained Language Models
Weihan Wang (Zhipu AI), Jie Tang (Tsinghua University)
TransformerLarge Language ModelVision Language ModelImageTextMultimodality
🎯 What it does: A visual expert module has been constructed, deeply integrating a frozen pre-trained language model with an image encoder to achieve a deep fusion of visual and language features.
Coherence-free Entrywise Estimation of Eigenvectors in Low-rank Signal-plus-noise Matrix Models
Hao Yan (University of Wisconsin Madison), Keith Levin (University of Wisconsin Madison)
🎯 What it does: A new algorithm for element-wise estimation of eigenvectors in a low-rank signal-noise matrix model is proposed; for the rank-one case, it is proven that the error is independent of the coherence of the matrix, and the optimal estimation rate (up to a logarithmic factor) is provided; an extended algorithm for the rank r case is given, which outperforms traditional spectral methods in experiments.
Coherent 3D Scene Diffusion From a Single RGB Image
Manuel Dahnert (Technical University of Munich), Matthias Nießner (Technical University of Munich)
GenerationPose EstimationDiffusion modelImage
🎯 What it does: A framework for 3D scene reconstruction from a single-view RGB image based on conditional diffusion is proposed, utilizing panoramic context to achieve globally consistent predictions of object pose and shape.
COLD: Causal reasOning in cLosed Daily activities
Abhinav Joshi (Indian Institute of Technology Kanpur), Ashutosh Modi (Indian Institute of Technology Kanpur)
Large Language ModelPrompt EngineeringText
🎯 What it does: This paper proposes the COLD framework, which constructs a closed causal graph using script-based knowledge of daily activities and automatically generates nearly ten million causal query triples through this graph to systematically evaluate the causal reasoning capabilities of large language models.
ColJailBreak: Collaborative Generation and Editing for Jailbreaking Text-to-Image Deep Generation
Yizhuo Ma (Xi'an Jiaotong University), Qing Guo (Agency for Science Technology and Research)
GenerationAdversarial AttackLarge Language ModelPrompt EngineeringDiffusion modelImageText
🎯 What it does: This paper proposes a collaborative generation and editing framework called ColJailBreak, which first generates safe images using secure natural language prompts, and then injects unsafe content through local editing to bypass the safety filters of commercial text-to-image models.
Collaboration! Towards Robust Neural Methods for Routing Problems
Jianan Zhou (Nanyang Technological University), Zhiqi Shen (Nanyang Technological University)
OptimizationAdversarial AttackTransformerReinforcement LearningTabular
🎯 What it does: A Collaborative Neural Framework (CNF) is proposed to enhance the performance of neural vehicle routing methods on clean and adversarial instances through multi-model collaborative adversarial training.
Collaborative Cognitive Diagnosis with Disentangled Representation Learning for Learner Modeling
Weibo Gao (University of Science and Technology of China), Zheng Zhang (University of Science and Technology of China)
Explainability and InterpretabilityRepresentation LearningGraph Neural NetworkAuto EncoderContrastive LearningTabular
🎯 What it does: This paper proposes a collaborative cognitive diagnosis model named Coral, which achieves interpretable diagnosis of learners' knowledge abilities by jointly learning intrinsic and collaborative information.
Collaborative Refining for Learning from Inaccurate Labels
BIN HAN, HUIMEI HE
ClassificationData-Centric LearningImageTextTabular
🎯 What it does: This paper proposes a collaborative refinement framework (CRL) that categorizes data into 'controversial' and 'fully consistent' based on annotator consistency, and effectively utilizes inaccurate labels through label refinement (LRD) and robust sample selection (RUS).
Collaborative Video Diffusion: Consistent Multi-video Generation with Camera Control
Zhengfei Kuang (Chinese University of Hong Kong), Gordon Wetzstein (Stanford University)
GenerationData SynthesisDiffusion modelVideo
🎯 What it does: This paper proposes a Collaborative Video Diffusion (CVD) framework that can generate high-quality videos with multiple consistent perspectives, synchronized content, and motion, given text prompts and multiple camera trajectories.
CoLoR-Filter: Conditional Loss Reduction Filtering for Targeted Language Model Pre-training
David Brandfonbrener (Kempner Institute at Harvard University), Sham M. Kakade
TransformerLarge Language ModelSupervised Fine-TuningText
🎯 What it does: This paper proposes CoLoR-Filter, an offline data selection method based on a two-model comparison, aimed at targeted pre-training of language models.
Color-Oriented Redundancy Reduction in Dataset Distillation
Bowen Yuan (University of Queensland), Zi Huang (University of Queensland)
Data SynthesisKnowledge DistillationImage
🎯 What it does: The AutoPalette framework is proposed to reduce color redundancy in synthesized images during dataset distillation through a palette network and color-guided initialization.
CoMat: Aligning Text-to-Image Diffusion Model with Image-to-Text Concept Matching
Dongzhi Jiang (Chinese University of Hong Kong), Hongsheng Li (Chinese University of Hong Kong)
Image TranslationGenerationVision Language ModelDiffusion modelImageTextMultimodality
🎯 What it does: This paper proposes an end-to-end fine-tuning strategy named CoMat, which enhances the alignment capability of text-to-image diffusion models by utilizing a concept matching mechanism from image to text.
Combining Observational Data and Language for Species Range Estimation
Max Hamilton (UMass Amherst), Subhransu Maji (UMass Amherst)
Large Language ModelText
🎯 What it does: Combining millions of citizen science observation data with Wikipedia text, the LE-SINR model is trained to achieve zero-shot prediction of species distribution ranges and improve accuracy in few-shot scenarios.
Combining Statistical Depth and Fermat Distance for Uncertainty Quantification
Hai-Vy Nguyen (Ampere Software Technology), Thierry Giaccone (Ampere Software Technology)
ClassificationAnomaly DetectionImage
🎯 What it does: This paper proposes a method to estimate out-of-distribution uncertainty by utilizing Lens Depth and an improved Fermat distance for deep measurement of samples in the feature space of pre-trained models, without the need for additional training or distribution assumptions.
CoMERA: Computing- and Memory-Efficient Training via Rank-Adaptive Tensor Optimization
Zi Yang (University at Albany), Zheng Zhang (University of California at Santa Barbara)
Recommendation SystemComputational EfficiencyTransformerTextTabular
🎯 What it does: The CoMERA framework is proposed, which implements a training method based on rank-adaptive tensor compression, significantly reducing memory usage while achieving speedup in training on GPUs.
Communication Bounds for the Distributed Experts Problem
Zhihao Jia (Carnegie Mellon University), Wenting Zheng (Carnegie Mellon University)
OptimizationHyperparameter SearchTabular
🎯 What it does: A low-communication online algorithm is designed for the distributed expert problem, achieving approximately optimal regret for aggregation functions such as summation, maximum, and ℓp norms under both message passing and broadcasting communication models.
Communication Efficient Distributed Training with Distributed Lion
Bo Liu (University of Texas at Austin), qiang liu
OptimizationComputational EfficiencyImageText
🎯 What it does: This paper proposes Distributed Lion, an improved version of the Lion optimizer that uses only binary or low-precision vector communication in distributed training, significantly reducing communication costs.
Communication-Efficient Federated Group Distributionally Robust Optimization
Zhishuai Guo (Texas A&M University), Tianbao Yang (Texas A&M University)
OptimizationFederated LearningTransformerLarge Language ModelImageText
🎯 What it does: This paper addresses the robust optimization problem caused by heterogeneous client data in federated learning and proposes three low-communication-complexity Federated Group Distributed Robust Optimization (FGDRO) algorithms: FGDRO-CVaR, FGDRO-KL, and FGDRO-KL-Adam.
Community Detection Guarantees using Embeddings Learned by Node2Vec
Andrew Davison (Columbia University), Owen G. Ward (Simon Fraser University)
Representation LearningGraph Neural NetworkGraph
🎯 What it does: This paper studies the theoretical properties of node2vec (and its DeepWalk variant) under the (degree-corrected) stochastic block model and proves that the learned embeddings can be used for weakly consistent community detection via k-means.
Compact Language Models via Pruning and Knowledge Distillation
Saurav Muralidharan (NVIDIA), Pavlo Molchanov (NVIDIA)
Computational EfficiencyKnowledge DistillationNeural Architecture SearchTransformerLarge Language ModelText
🎯 What it does: The paper studies how to quickly obtain a smaller high-performance LLM by combining structured pruning (width and depth) with knowledge distillation, using a small amount of data from the original large model.
Compact Proofs of Model Performance via Mechanistic Interpretability
Jason Gross (Massachusetts Institute of Technology), Lawrence Chan
Explainability and InterpretabilityTransformerTabular
🎯 What it does: Using mechanism interpretation to perform reverse analysis on a single-layer single-head attention Transformer executing the Max-of-K task, various forms of formal proofs are constructed based on the obtained implementation details, and a lower bound on the model's accuracy is calculated.
Complete Graphical Criterion for Sequential Covariate Adjustment in Causal Inference
Yonghan Jung (Purdue University), Sanghack Lee (Seoul National University)
Graph
🎯 What it does: This paper proposes a new graphical discrimination criterion - the Sequential Adjustment Criterion (SAC), used to identify causal effects that can be adjusted through covariates in sequential causal models.
Compositional 3D-aware Video Generation with LLM Director
Hanxin Zhu (University of Science and Technology of China), Jiang Bian (Microsoft Research Asia)
GenerationData SynthesisPose EstimationTransformerLarge Language ModelDiffusion modelScore-based ModelGaussian SplattingVideoMultimodality
🎯 What it does: A C3V framework based on LLM directing is proposed, which first decomposes the text into concepts such as scenes, objects, and actions. It generates 3D representations using 3D Gaussian spheres and a motion library, and then refines trajectories and poses through a multimodal LLM and Score Distillation from a 2D diffusion model, achieving high-quality and controllable 3D video generation.
Compositional Automata Embeddings for Goal-Conditioned Reinforcement Learning
Beyazit Yalcinkaya (University of California), Sanjit A. Seshia (University of California)
Graph Neural NetworkReinforcement LearningGraph
🎯 What it does: A target condition reinforcement learning framework based on combinatorial deterministic finite automata (cDFA) has been designed and implemented, using Graph Attention Networks (GATv2) to embed the cDFA and pre-trained on the Reach-Avoid Derivative (RAD) automaton to achieve zero-shot generalization and accelerated learning for unknown tasks.
Compositional Generalization Across Distributional Shifts with Sparse Tree Operations
Paul Soulos (Johns Hopkins University), Roland Fernandez (Microsoft Research)
TransformerTextSequential
🎯 What it does: Proposes a Sparse Differentiable Tree Machine (sDTM) and uses Sparse Coordinate Trees (SCT) to represent tree structures, unifying neural and symbolic computation.
Compositional PAC-Bayes: Generalization of GNNs with persistence and beyond
Kirill Brilliantov (ETH Zurich), Vikas Garg (Aalto University)
Graph Neural NetworkGraph
🎯 What it does: This paper proposes a composable PAC-Bayes framework that provides data-dependent consistency generalization bounds for heterogeneous layers and sub-models (such as graph neural networks and persistent layers like PersLay), and designs new regularization methods based on these bounds.
Compressing Large Language Models using Low Rank and Low Precision Decomposition
Rajarshi Saha (Stanford University), Mert Pilanci (Stanford University)
CompressionTransformerLarge Language ModelText
🎯 What it does: The CALDERA algorithm is proposed, which performs low-precision low-rank decomposition (Q + LR) on LLM weight matrices and supports post-training quantization and low-rank adaptation.
Computation-Aware Gaussian Processes: Model Selection And Linear-Time Inference
Jonathan Wenger (Columbia University), John Patrick Cunningham
OptimizationComputational EfficiencyTabular
🎯 What it does: This paper proposes a linear time model selection and inference method for Computationally Aware Gaussian Processes (CaGP), addressing the computational and memory bottlenecks of traditional Gaussian Processes in large-scale data.
Computational Aspects of Bayesian Persuasion under Approximate Best Response
Kunhe Yang (University of California), Hanrui Zhang (Chinese University of Hong Kong)
Optimization
🎯 What it does: This paper studies the robust Bayesian persuasion problem with approximately optimal responses from the receiver and proposes various algorithms for different parameter scales (finite state space or action space); it also proves that the problem is NP-hard in general and provides a quasi-polynomial time approximation scheme (QPTAS).
Computerized Adaptive Testing via Collaborative Ranking
Zirui Liu (University of Science and Technology of China), Shijin Wang (iFLYTEK Co., Ltd)
Tabular
🎯 What it does: This paper proposes a Collaborative Computer Adaptive Testing (CCAT) framework, which utilizes collaborative students who have completed the entire question bank as a benchmark to improve question selection and ability estimation, thereby enhancing the consistency of student rankings.
Computing the Bias of Constant-step Stochastic Approximation with Markovian Noise
Sebastian Allmeier (University of Grenoble Alpes and Inria), Nicolas Gast (University of Grenoble Alpes and Inria)
Stochastic Differential Equation
🎯 What it does: This paper studies the bias of constant step-size stochastic approximation algorithms under Markov noise and provides precise order estimates.
Con4m: Context-aware Consistency Learning Framework for Segmented Time Series Classification
Junru Chen (Zhejiang University), Yang Yang (Zhejiang University)
ClassificationSegmentationTransformerTime Series
🎯 What it does: A segmented classification framework called Con4m is proposed for original multi-class time series with varying durations (MVD), utilizing contextual information to enhance segmentation classification performance and gradually harmonizing inconsistent boundary labels.
Concentrate Attention: Towards Domain-Generalizable Prompt Optimization for Language Models
Chengzhengxu Li (Xi'an Jiaotong University), Chao Shen (Xi'an Jiaotong University)
Domain AdaptationOptimizationTransformerLarge Language ModelReinforcement LearningPrompt EngineeringText
🎯 What it does: A prompt optimization objective based on attention concentration is proposed, with new loss and matching strategies designed for both soft prompts and hard prompts, thereby enhancing the generalization performance of language models in unknown domains.
Conditional Controllable Image Fusion
Bing Cao (Tianjin University), Qinghua Hu (Tianjin University)
Image TranslationData SynthesisDiffusion modelImageMultimodalityMagnetic Resonance Imaging
🎯 What it does: A conditional controllable image fusion framework (CCF) based on a pre-trained denoising diffusion model is proposed, achieving training-free fusion through a conditional library and sampling adaptive condition selection.
Conditional Density Estimation with Histogram Trees
Lincen Yang (Leiden University), Matthijs van Leeuwen (Leiden University)
Tabular
🎯 What it does: A Conditional Density Tree (CDTree) is designed and implemented, which partitions the feature space using decision trees. Each leaf approximates the conditional density using histograms, and the tree structure and leaf histogram parameters are learned uniformly through the Minimum Description Length (MDL) principle.
Conditional Generative Models are Sufficient to Sample from Any Causal Effect Estimand
Md Musfiqur Rahman (Purdue University), Murat Kocaoglu (Purdue University)
GenerationData SynthesisDiffusion modelGenerative Adversarial NetworkImageTextBiomedical DataMagnetic Resonance Imaging
🎯 What it does: An algorithm ID-GEN based on conditional generative models is proposed, which can perform high-dimensional sampling of any identifiable causal effect distribution (observable or interventional) in the presence of unobserved confounding factors;
Conditional Outcome Equivalence: A Quantile Alternative to CATE
Josh Givens (University of Bristol), Katarzyna Reluga (University of Bristol)
TabularTime Series
🎯 What it does: A Conditional Quantile Comparator (CQC) is proposed as a new indicator for heterogeneous treatment effects, along with an estimation method based on doubly robust pseudo-outcomes.
Conditional Synthesis of 3D Molecules with Time Correction Sampler
Hojung Jung (KAIST AI), Jinwoo Shin (KAIST AI)
GenerationData SynthesisDrug DiscoveryGraph Neural NetworkDiffusion modelGraph
🎯 What it does: This paper proposes a time-aware conditional synthesis (TACS) framework based on diffusion models, aimed at simultaneously satisfying target attributes while maintaining molecular validity and stability during 3D molecular generation.
Conditioning non-linear and infinite-dimensional diffusion processes
Elizabeth Louise Baker (University of Copenhagen), Stefan Sommer (University of Copenhagen)
Time SeriesStochastic Differential Equation
🎯 What it does: This paper proposes a method for conditioning nonlinear diffusion processes in infinite-dimensional spaces, utilizing infinite-dimensional Doob-h transformations and fractional matching to achieve precise or approximate conditioning of function value stochastic processes.
CondTSF: One-line Plugin of Dataset Condensation for Time Series Forecasting
Jianrong Ding (Shanghai Jiao Tong University), Linghe Kong (Shanghai Jiao Tong University)
Data SynthesisOptimizationKnowledge DistillationTime Series
🎯 What it does: A plugin named CondTSF is proposed to improve dataset distillation in time series forecasting, primarily by optimizing the value term to enhance the quality of synthetic data.
Confidence Calibration of Classifiers with Many Classes
Adrien Le Coz (IRT SystemX), Faouzi Adjed (IRT SystemX)
ClassificationSupervised Fine-TuningImageText
🎯 What it does: A method is proposed to transform the confidence calibration problem of multi-class classifiers into a single binary classification problem—Top-versus-All (TvA), and improvements to existing calibration methods are made within this framework;
Confidence Regulation Neurons in Language Models
Alessandro Stolfo (ETH Zurich), Neel Nanda
GenerationExplainability and InterpretabilityTransformerLarge Language ModelText
🎯 What it does: This study investigates the confidence modulation mechanisms within large-scale language models (LLMs) and identifies two types of neurons: entropy neurons, which adjust output entropy by writing into the null space of unembedding and utilizing LayerNorm; and token frequency neurons, which directly adjust the distance between the output distribution and the unigram distribution, thereby affecting the model's confidence. Their functionality is validated in the context of repeated sequences (induction).
Confident Natural Policy Gradient for Local Planning in $q_\pi$-realizable Constrained MDPs
Tian Tian (University of Alberta), Csaba Szepesvari
OptimizationReinforcement Learning
🎯 What it does: A CMDP planning algorithm based on primal-dual and natural policy gradient is proposed, which can output an approximately optimal policy that satisfies relaxed or strict constraints with high probability in an infinite state space with qπ-achievable linear function approximation under a locally accessible model.
Conformal Alignment: Knowing When to Trust Foundation Models with Guarantees
Yu Gui (University of Chicago), Zhimei Ren (University of Pennsylvania)
GenerationData-Centric LearningSupervised Fine-TuningTextBiomedical Data
🎯 What it does: To address the alignment issue of large foundational model outputs, we propose the Conformal Alignment framework, which utilizes distribution-independent conformal prediction to select individual outputs, ensuring that the selected outputs are highly consistent with human evaluations under a given FDR target.
Conformal Classification with Equalized Coverage for Adaptively Selected Groups
Yanfei Zhou (University of Southern California), Matteo Sesia (University of Southern California)
ClassificationTabularBiomedical Data
🎯 What it does: The AFCP method is proposed, which utilizes adaptive selection of sensitive attributes and constructs a classification prediction set that satisfies adaptive equalized coverage.
Conformal Inverse Optimization
Bo Lin (University of Toronto), Timothy Chan
OptimizationTabular
🎯 What it does: A new inverse optimization process is proposed, which first learns the uncertainty set from decision data and then provides new decisions using robust optimization.
Conformal Prediction for Class-wise Coverage via Augmented Label Rank Calibration
Yuanjie Shi (Washington State University), Yan Yan (Washington State University)
ClassificationComputational EfficiencyImage
🎯 What it does: The RC3P algorithm is proposed, improving the traditional class-conditional CP, resulting in a significantly smaller prediction set while ensuring coverage for each category.
Conformalized Credal Set Predictors
Alireza Javanmardi (Ludwig Maximilian University of Munich), Eyke Hüllermeier (Ludwig Maximilian University of Munich)
ClassificationImageText
🎯 What it does: This paper proposes a credible set prediction method based on conformal prediction to provide a set that includes the true distribution in classification tasks and quantifies randomness and knowledge uncertainty.
Conformalized Multiple Testing after Data-dependent Selection
Xiaoning Wang (Nankai University), Changliang Zou (Nankai University)
Tabular
🎯 What it does: This study investigates how to achieve FDR control in multiple testing methods by constructing selective conformal p-values and combining them with the BH procedure after data-driven selection.
Conformalized Time Series with Semantic Features
Baiting Chen (University of California Los Angeles), Lu Cheng (University of Illinois)
Recurrent Neural NetworkTransformerTime SeriesFinance Related
🎯 What it does: This paper proposes the CT-SSF (Conformalized Time Series with Semantic Features) method, which constructs non-compliance scores in the semantic latent space of neural networks and dynamically adjusts weights to achieve confidence interval prediction for time series.
Confusion-Resistant Federated Learning via Diffusion-Based Data Harmonization on Non-IID Data
xiaohong chen, Yongmei liu
Federated LearningDiffusion modelImage
🎯 What it does: Proposes the CRFed framework to address the model inconsistency problem caused by non-IID data in FL.
Conjugate Bayesian Two-step Change Point Detection for Hawkes Process
Zeyue Zhang (Renmin University of China), Feng Zhou (Renmin University of China)
Time Series
🎯 What it does: A conjugate Bayesian two-step change point detection method CoBay-CPD is proposed to capture abrupt changes in the parameters of the Hawkes process over time;
Conjugated Semantic Pool Improves OOD Detection with Pre-trained Vision-Language Models
Mengyuan Chen (Institute of Automation, Chinese Academy of Sciences), Changsheng Xu (Institute of Automation, Chinese Academy of Sciences)
Anomaly DetectionTransformerVision Language ModelImageText
🎯 What it does: A theoretical framework for zero-shot OOD detection is proposed, and based on this, a conjugated semantic pool (CSP) is designed to enhance performance.
Connecting Joint-Embedding Predictive Architecture with Contrastive Self-supervised Learning
Shentong Mo (Carnegie Mellon University), Shengbang Tong (New York University)
Object DetectionSegmentationRepresentation LearningTransformerContrastive LearningImageVideo
🎯 What it does: A framework for unsupervised visual representation learning called C-JEPA is proposed, which combines JEPA and VICReg to enhance representation quality through mask prediction and contrastive regularization.
Connecting the Dots: LLMs can Infer and Verbalize Latent Structure from Disparate Training Data
Johannes Treutlein (University of California Berkeley), Owain Evans
TransformerLarge Language ModelSupervised Fine-TuningText
🎯 What it does: This study explores whether large language models (LLMs) can infer hidden facts (referred to as inductive out-of-context reasoning, OOCR) by aggregating dispersed, implicit information from training data, and validates their 'connection point' ability through fine-tuning the model and evaluating it on five different tasks.
Connectivity Shapes Implicit Regularization in Matrix Factorization Models for Matrix Completion
Zhiwei Bai (Shanghai Jiao Tong University), Yaoyu Zhang (Shanghai Jiao Tong University)
Graph
🎯 What it does: This study investigates the implicit regularization of matrix decomposition models in matrix completion, exploring how the connectivity of observed data determines the preference for low-rank or low nuclear norm.
Connectivity-Driven Pseudo-Labeling Makes Stronger Cross-Domain Segmenters
Dong Zhao (Xidian University), Zhun Zhong (University of Nottingham)
SegmentationDomain AdaptationImage
🎯 What it does: A pseudo-label generation framework based on semantic connectivity (SeCo) is proposed, which aggregates noisy pixels into connected regions for denoising and correction to enhance cross-domain semantic segmentation performance.
Consensus Learning with Deep Sets for Essential Matrix Estimation
Dror Moran (Weizmann Institute of Science), Ronen Basri (Weizmann Institute of Science)
OptimizationComputational EfficiencyImage
🎯 What it does: A NACNet network based on Deep Sets has been designed and implemented for consistent learning from matching sets containing a large number of outliers and noise, first denoising inliers and then classifying, ultimately regressing the essential matrix through weighted DLT.
Consistency Diffusion Bridge Models
Guande He (Tsinghua University), Jun Zhu (Tsinghua University)
Image TranslationRestorationGenerationData SynthesisComputational EfficiencyKnowledge DistillationDiffusion modelImageOrdinary Differential Equation
🎯 What it does: This paper proposes a Consistency Diffusion Bridge Model (CDBM), which extends the Consistency Models to the Denoising Diffusion Bridge Model (DDBM), significantly improving the sampling efficiency of DDBM while maintaining or enhancing the generation quality.
Consistency Models for Scalable and Fast Simulation-Based Inference
Marvin Schmitt (University of Stuttgart), Stefan T. Radev (Rensselaer Polytechnic Institute)
RestorationOptimizationComputational EfficiencyConvolutional Neural NetworkFlow-based ModelImage
🎯 What it does: This paper proposes a posterior estimation method based on a consistency model (CMPE) for efficient and scalable simulation-driven inference.
Consistency of Neural Causal Partial Identification
Jiyuan Tan (Stanford University), Vasilis Syrgkanis (Stanford University)
🎯 What it does: A partial identification method based on Neural Causal Models (NCM) is proposed, and its consistency is proven under a general Structural Causal Model (SCM) that includes both continuous and categorical variables.
Consistency Purification: Effective and Efficient Diffusion Purification towards Certified Robustness
Yiquan Li (University of Wisconsin-Madison), Chaowei Xiao (University of Wisconsin-Madison)
RestorationGenerationOptimizationDiffusion modelImage
🎯 What it does: A single-step diffusion purification framework called Consistency Purification is proposed, utilizing a consistency model and further fine-tuned with LPIPS loss to enhance semantic alignment.
Constant Acceleration Flow
Dogyun Park (Korea University), Hyunwoo J. Kim (Korea University)
GenerationFlow-based ModelImageOrdinary Differential Equation
🎯 What it does: By introducing a learnable acceleration term, we propose Constant Acceleration Flow (CAF) to approximate ODE flows, improving the quality of single sampling.
ConStat: Performance-Based Contamination Detection in Large Language Models
Jasper Dekoninck (ETH Zurich), Martin Vechev (ETH Zurich)
TransformerLarge Language ModelTextBenchmark
🎯 What it does: A performance-based detection method called CONSTAT has been designed to identify performance exaggeration caused by data contamination in benchmark tests of large language models.
Constrained Adaptive Attack: Effective Adversarial Attack Against Deep Neural Networks for Tabular Data
Thibault Simonetto (University of Luxembourg), Maxime Cordy (University of Luxembourg)
Adversarial AttackTabular
🎯 What it does: This paper proposes two new adversarial attack methods, CAPGD (Non-parametric Adaptive Projected Gradient Descent) and CAA (a global attack that combines CAPGD with the search attack MOEVA), to evaluate the robustness of deep learning models on tabular data while satisfying constraints of category, value, boundary, and feature relationships.
Constrained Binary Decision Making
Daniel Průša (Czech Technical University in Prague), Vojtech Franc (Czech Technical University in Prague)
Optimization
🎯 What it does: A generalized binary decision-making framework is proposed, providing a complete optimal strategy theory and demonstrating how to directly solve optimal choice strategies for various known BDM tasks (Neyman-Pearson, Bounded-Abstention, Bounded-Improvement, SCOD, etc.).
Constrained Diffusion Models via Dual Training
Shervin Khalafi (University of Pennsylvania), Alejandro Ribeiro (University of Pennsylvania)
GenerationData SynthesisOptimizationDiffusion modelImage
🎯 What it does: This paper proposes a framework for training diffusion models under distribution constraints and implements dual training through the Lagrange dual method to meet specified probability distribution requirements.
Constrained Diffusion with Trust Sampling
William Huang (Stanford University), Karen Liu
GenerationSuper ResolutionDiffusion modelImageVideo
🎯 What it does: A novel training-free guided diffusion method called Trust Sampling is proposed, which satisfies constraints through multi-step gradient optimization at each diffusion step.
Constrained Latent Action Policies for Model-Based Offline Reinforcement Learning
Marvin Alles (Volkswagen Group), Maximilian Karl (Volkswagen Group)
Reinforcement LearningTabularSequential
🎯 What it does: A model-based offline reinforcement learning method called C-LAP is proposed, which generates actions in the latent action space and maintains the policy within the dataset distribution through support constraints of the latent action distribution, addressing the value overestimation problem caused by model errors.
Constrained Sampling with Primal-Dual Langevin Monte Carlo
Luiz F. O. Chamon (University of Stuttgart), Anna Korba (CREST, ENSAE, IP Paris)
Anomaly DetectionOptimizationTabularTime SeriesFinance RelatedStochastic Differential Equation
🎯 What it does: This paper studies the problem of sampling probability distributions under statistical constraints (such as expectation constraints) and proposes a Primal-Dual Langevin Monte Carlo (PD-LMC) algorithm without explicit integration;
Constrained Synthesis with Projected Diffusion Models
Jacob K Christopher, Ferdinando Fioretto (University of Virginia)
GenerationData SynthesisOptimizationDiffusion modelScore-based ModelImageVideo
🎯 What it does: This paper proposes a framework called Projected Diffusion Models (PDM), which utilizes projection iteration to treat the reverse process of diffusion models as a constrained optimization problem, thereby achieving strict satisfaction of arbitrary constraints (including convex, non-convex, and ODE physical constraints) when generating samples.
Constructing Semantics-Aware Adversarial Examples with a Probabilistic Perspective
Andi Zhang (Computer Laboratory University of Cambridge), Damon Wischik (Computer Laboratory University of Cambridge)
Adversarial AttackDiffusion modelImage
🎯 What it does: This paper proposes a method for generating semantically aware adversarial examples from a probabilistic perspective.
Construction and Application of Materials Knowledge Graph in Multidisciplinary Materials Science via Large Language Model
Yanpeng Ye (University of New South Wales), Wenjie Zhang (University of New South Wales)
Large Language ModelTextGraph
🎯 What it does: A material knowledge graph (MKG) was constructed and applied, utilizing large language models to extract and standardize entities and relationships from 150,000 abstracts of energy materials literature, and link prediction was achieved through graph embedding and network algorithms.
ContactField: Implicit Field Representation for Multi-Person Interaction Geometry
Hansol Lee (Korea Institute of Science and Technology), Hwasup Lim (Korea Institute of Science and Technology)
Data SynthesisPose EstimationTransformerContrastive LearningPoint CloudMesh
🎯 What it does: This paper proposes an implicit field representation that can simultaneously estimate the occupancy rate, identity labels (ID), and contact fields of multiple human interactions, achieving precise reconstruction of occlusions and close-range interactions through a multi-view local-global feature module.
Context and Geometry Aware Voxel Transformer for Semantic Scene Completion
Zhu Yu (Zhejiang University), Hui-liang Shen
SegmentationAutonomous DrivingTransformerPoint Cloud
🎯 What it does: A context and geometry-aware voxel transformer called CGFormer is proposed for camera-aware semantic scene completion.
Context-Aware Testing: A New Paradigm for Model Testing with Large Language Models
Paulius Rauba (University of Cambridge), Mihaela van der Schaar (University of Cambridge)
TransformerLarge Language ModelTabularBiomedical DataFinance Related
🎯 What it does: A Context-Aware Testing (CAT) framework is proposed, and the SMART Testing system is implemented, which uses large language models (LLMs) to generate potential model failure hypotheses, automate verification, and generate reports.
ContextCite: Attributing Model Generation to Context
Benjamin Cohen-Wang (Massachusetts Institute of Technology), Aleksander Madry (Massachusetts Institute of Technology)
Explainability and InterpretabilityAdversarial AttackTransformerLarge Language ModelText
🎯 What it does: A new context attribution method called CONTEXTCITE is proposed and implemented to determine the context fragments that the language model actually relies on when generating specific statements.
ContextGS : Compact 3D Gaussian Splatting with Anchor Level Context Model
Yufei Wang (Nanyang Technological University), Bihan Wen (Nanyang Technological University)
CompressionGaussian SplattingPoint Cloud
🎯 What it does: A compression framework based on 3D Gaussian Splatting (ContextGS) is proposed, utilizing a hierarchical anchor autoregressive model and hyper-prior features for efficient encoding.
Contextual Active Model Selection
Xuefeng Liu (University of Chicago), Yuxin Chen (University of Chicago)
ClassificationOptimizationMultimodality
🎯 What it does: This paper proposes an online context-aware active model selection framework called CAMS, which can adaptively select the optimal model from pre-trained models and learn under limited label costs.
Contextual Bilevel Reinforcement Learning for Incentive Alignment
Vinzenz Thoma (ETH AI Center), Yifan Hu (ETH Zurich)
OptimizationReinforcement LearningTabular
🎯 What it does: This paper proposes the Contextual Bilevel Reinforcement Learning (CB-RL) framework, which studies the bilevel MDP problem in stochastic contexts and multi-follower environments, and presents a trajectory-based stochastic hypergradient descent algorithm (HPGD), further providing an accelerated version of RT-Q for controllable lower-level learning.
Contextual Decision-Making with Knapsacks Beyond the Worst Case
Zhaohua Chen (Peking University), Xiaotie Deng (Peking University)
OptimizationReinforcement Learning from Human FeedbackReinforcement Learning
🎯 What it does: The study focuses on contextual decision-making and the knapsack constraint problem, proposing a low-scheduling error algorithm under non-degenerate fluid linear programming (LP) conditions, and providing a lower bound under degenerate conditions.
Contextual Linear Optimization with Bandit Feedback
Yichun Hu (Cornell University), Yanchen Wu (Tsinghua University)
OptimizationReinforcement LearningTabular
🎯 What it does: This paper studies the contextual linear optimization (CLO) problem under bandit feedback, where only decision costs can be observed, and proposes an end-to-end Induced Empirical Risk Minimization (IERM) algorithm.
Contextual Multinomial Logit Bandits with General Value Functions
Mengxiao Zhang (University of Iowa), Haipeng Luo (University of Southern California)
Recommendation SystemOptimizationComputational EfficiencyTabular
🎯 What it does: This paper studies context-aware multinomial logit (MNL) slot machines with general value functions and proposes a series of algorithms to address recommendation problems in both stochastic and adversarial environments.
Continual Audio-Visual Sound Separation
Weiguo Pian (University of Texas at Dallas), Yapeng Tian (University of Texas at Dallas)
Knowledge DistillationTransformerContrastive LearningMultimodalityAudio
🎯 What it does: Proposes a continuous audio-visual sound separation task and designs the ContAV-Sep framework, allowing the model to maintain separation performance for old categories while learning new sound source categories.
Continual Counting with Gradual Privacy Expiration
Joel Daniel Andersson (Copenhagen University), Jalaj Upadhyay (Rutgers University)
Safty and PrivacyTabular
🎯 What it does: This study investigates the introduction of an asymptotic privacy decay model in continuous counting problems, providing algorithms that satisfy different privacy decay functions and proving the matching lower bound of the error with respect to privacy decay.
Continual Learning in the Frequency Domain
RuiQi Liu, Yongjun Xu (Institute of Computing Technology Chinese Academy of Sciences)
ClassificationOptimizationComputational EfficiencyConvolutional Neural NetworkSupervised Fine-TuningImage
🎯 What it does: This paper proposes a framework for incremental learning in the frequency domain called CLFD, which improves the efficiency and accuracy of continuous learning by utilizing frequency domain feature encoding and class-aware feature selection.
Continual Learning with Global Alignment
Xueying Bai (Stony Brook University), Niranjan Balasubramanian (Stony Brook University)
TransformerSupervised Fine-TuningText
🎯 What it does: A global alignment method is proposed, utilizing pre-trained word representations as a foundation, and learning data representations through task-specific interpolation or low-rank adaptation (LoRA), thereby reducing cross-task gradient interference and catastrophic forgetting during continual learning.
Continual learning with the neural tangent ensemble
Ari S Benjamin, Kyle Daruwalla (Cold Spring Harbor Laboratory)
Convolutional Neural NetworkMixture of ExpertsImage
🎯 What it does: Proposes to view a single neural network as a neural tangent ensemble (NTE) of parameter-level experts, and utilizes its posterior updates to achieve continual learning, avoiding catastrophic forgetting;