IJCAI 2023 Papers — Page 4
International Joint Conference on Artificial Intelligence · 639 papers
Incomplete Multi-view Clustering via Prototype-based Imputation
Haobin Li (Sichuan University), Xi Peng (Sichuan University)
Representation LearningData-Centric LearningContrastive LearningMultimodality
🎯 What it does: Studied incomplete multi-view clustering, proposing a prototype-oriented imputation method to preserve instance commonality and view diversity.
Incorporating Unlikely Negative Cues for Distinctive Image Captioning
Zhengcong Fei (Meituan), Junshi Huang (Meituan)
GenerationKnowledge DistillationTransformerVision Language ModelImageTextMultimodalityRetrieval-Augmented Generation
🎯 What it does: Propose a framework named UNKT that leverages negative knowledge distillation to enhance the diversity of image captions.
Incremental and Decremental Optimal Margin Distribution Learning
Li-Jun Chen (Huazhong University of Science and Technology), Hai Jin (Huazhong University of Science and Technology)
ClassificationOptimizationComputational EfficiencyTabular
🎯 What it does: Propose an ID-ODM algorithm that simultaneously handles incremental and decremental learning, enabling efficient updates of the optimal marginal distribution machine (ODM) model when data streams or privacy protection render old data invalid.
Independent Feature Decomposition and Instance Alignment for Unsupervised Domain Adaptation
Qichen He (University of Electronic Science and Technology of China), Dongde Hou (Southwest University of Political Science & Law)
Domain AdaptationFlow-based ModelContrastive LearningImageBenchmark
🎯 What it does: Proposes an independent feature decomposition and instance alignment framework (IndUDA) based on invertible flow, which maps features to a latent space and separates domain-invariant and domain-specific dimensions, achieving domain-invariant feature adaptation through domain-internal exchange and noise replacement;
Inducing Stackelberg Equilibrium through Spatio-Temporal Sequential Decision-Making in Multi-Agent Reinforcement Learning
Bin Zhang (Institute of Automation Chinese Academy of Sciences), Guoliang Fan (Institute of Automation Chinese Academy of Sciences)
Reinforcement LearningSequentialBenchmark
🎯 What it does: Proposed a multi-agent reinforcement learning method called STEP based on the Stackelberg leader model, which constructs a spatiotemporal sequence Markov game (STMG) during training and employs a conditional hypernetwork to generate N-level strategy models during execution, achieving asynchronous action coordination and ultimately converging to the Stackelberg equilibrium.
Inferring Private Valuations from Behavioral Data in Bilateral Sequential Bargaining
Lvye Cui (Beijing Institute of Technology), Haoran Yu (Beijing Institute of Technology)
Recurrent Neural NetworkTabularSequentialFinance Related
🎯 What it does: This paper proposes a Bayesian learning-based private valuation inference framework called BLUE, which infers sellers' private valuations of goods from bilateral sequential bargaining behaviors under the assumption of non-strictly dominant strategies.
InitLight: Initial Model Generation for Traffic Signal Control Using Adversarial Inverse Reinforcement Learning
Yutong Ye (East China Normal University), Xiang Lian (Kent State University)
Autonomous DrivingReinforcement LearningGenerative Adversarial NetworkGraph
🎯 What it does: Proposed InitLight, a pre-training method based on adversarial inverse reinforcement learning (AIRL), which rapidly generates a general traffic signal control (TSC) initial model using expert trajectories from a single intersection environment.
Intent-aware Recommendation via Disentangled Graph Contrastive Learning
Yuling Wang (Beijing University of Posts and Telecommunications), Wei Wu (Meituan)
Recommendation SystemExplainability and InterpretabilityGraph Neural NetworkContrastive LearningTabular
🎯 What it does: Propose an intent-aware recommendation model IDCL based on graph neural networks, which can simultaneously learn interpretable user intentions and their behavior distributions.
Invertible Residual Neural Networks with Conditional Injector and Interpolator for Point Cloud Upsampling
Aihua Mao (South China University of Technology), Yong-Jin Liu (Tsinghua University)
Super ResolutionFlow-based ModelPoint Cloud
🎯 What it does: This paper proposes a point cloud up-sampling method called PU-INN based on invertible residual networks.
iRe2f: Rethinking Effective Refinement in Language Structure Prediction via Efficient Iterative Retrospecting and Reasoning
Zuchao Li (Wuhan University), Hai Zhao (Shanghai Jiao Tong University)
Computational EfficiencyRepresentation LearningGraph Neural NetworkTextGraph
🎯 What it does: This paper proposes a lightweight iterative revision framework, IRE F, which revises language structure predictions by leveraging retrospection and reasoning on graph representations without re-encoding.
JEPOO: Highly Accurate Joint Estimation of Pitch, Onset and Offset for Music Information Retrieval
Haojie Wei (Renmin University of China), Gang Wang (Huawei)
RetrievalConvolutional Neural NetworkRecurrent Neural NetworkAudio
🎯 What it does: Propose a joint pitch, onset, and offset detection method named JEPOO, which can simultaneously process both single-pitch and multi-pitch musical data;
Joint-MAE: 2D-3D Joint Masked Autoencoders for 3D Point Cloud Pre-training
Ziyu Guo (Chinese University of Hong Kong), Pheng-Ann Heng (Chinese University of Hong Kong)
Representation LearningTransformerAuto EncoderImagePoint Cloud
🎯 What it does: Propose the Joint-MAE framework, which performs joint masked autoencoding using 3D point clouds and their projected 2D depth maps to enhance 3D point cloud pre-training effectiveness.
K∗ Search over Orbit Space for Top-k Planning
Michael Katz (IBM T.J. Watson Research Center), Junkyu Lee (IBM T.J. Watson Research Center)
OptimizationBenchmark
🎯 What it does: Propose a new algorithm called OK* that performs K* search in the orbit space, achieving efficient solutions for Top-k planning problems.
KDLGT: A Linear Graph Transformer Framework via Kernel Decomposition Approach
Yi Wu (Peking University), Shaoguo Liu (Alibaba Group)
Computational EfficiencyGraph Neural NetworkTransformerGraph
🎯 What it does: Propose the KDLGT framework, reducing the quadratic complexity of GT to linear, and design two instance models, LSAT and SAPDGT.
Keep Skills in Mind: Understanding and Implementing Skills in Commonsense Question Answering
Meikai Bao (University of Science and Technology of China), Jun Zhou (University of Science and Technology of China)
Explainability and InterpretabilityTransformerLarge Language ModelPrompt EngineeringText
🎯 What it does: This paper proposes a dynamic skill-aware closed-source commonsense question answering framework, DSCQA, which enhances the model's reasoning ability by extracting skill features from the training set and dynamically injecting skill information during decoding.
KEST: Kernel Distance Based Efficient Self-Training for Improving Controllable Text Generation
Yuxi Feng (University of British Columbia), Xing Xie (Microsoft Research Asia)
GenerationTransformerLarge Language ModelText
🎯 What it does: This paper proposes an efficient self-training framework called KEST based on kernel distance, aiming to improve the control accuracy and diversity of controllable text generation while significantly accelerating the pseudo-text generation process.
KMF: Knowledge-Aware Multi-Faceted Representation Learning for Zero-Shot Node Classification
Likang Wu (University of Science and Technology of China), Enhong Chen (University of Science and Technology of China)
ClassificationRepresentation LearningGraph Neural NetworkLarge Language ModelContrastive LearningTextGraph
🎯 What it does: Proposed a knowledge graph-based multi-dimensional representation learning framework called KMF for zero-shot node classification.
Label Enhancement via Joint Implicit Representation Clustering
Yunan Lu (Nanjing University of Science and Technology), Xiuyi Jia (Nanjing University of Science and Technology)
Representation LearningGraph Neural NetworkAuto EncoderImageText
🎯 What it does: Propose deep generative models JRC and LEIC, which jointly learn implicit representations of features and simple labels through clustering to enhance label augmentation effectiveness.
Label Specific Multi-Semantics Metric Learning for Multi-Label Classification: Global Consideration Helps
Jun-Xiang Mao (Southeast University), Min-Ling Zhang (Southeast University)
ClassificationImageTextBenchmark
🎯 What it does: Propose a multi-semantic metric learning framework (LIMIC) that simultaneously learns global and multi-label specific local metrics for multi-label classification.
Latent Processes Identification From Multi-View Time Series
Zenan Huang (Zhejiang University), Nenggan Zheng (Zhejiang University)
Explainability and InterpretabilityRepresentation LearningContrastive LearningTime Series
🎯 What it does: Propose a framework named MuLTI for identifying identifiable causal latent variables from multi-perspective time series, and merge the latent variables extracted from each perspective into a complete latent variable through optimal transport.
Learnable Surrogate Gradient for Direct Training Spiking Neural Networks
Shuang Lian (Zhejiang University), Huajin Tang (Zhejiang University)
ClassificationSpiking Neural NetworkImage
🎯 What it does: Propose a learnable alternative gradient (LSG) method that uses a trainable decay factor to dynamically adjust the gradient width near the activation threshold, enabling direct training of deep spiking neural networks (SNNs)
Learning 3D Photography Videos via Self-supervised Diffusion on Single Images
Xiaodong Wang (Peking University), Nan Duan (Microsoft)
GenerationData SynthesisDepth EstimationVision Language ModelDiffusion modelImageText
🎯 What it does: Designed and implemented a method for generating 3D photorealistic videos from a single image based on self-supervised diffusion models, and proposed a new task called out-animation that extends input objects in space and time.
Learning Attention from Attention: Efficient Self-Refinement Transformer for Face Super-Resolution
Guanxin Li (Xi'an Jiaotong University), Yihong Gong (Xi'an Jiaotong University)
Super ResolutionTransformerImage
🎯 What it does: Proposed a self-refinement Transformer framework for facial image super-resolution tasks;
Learning Calibrated Uncertainties for Domain Shift: A Distributionally Robust Learning Approach
Haoxuan Wang (Shanghai Jiao Tong University), Junchi Yan (Shanghai Jiao Tong University)
Domain AdaptationImage
🎯 What it does: Propose an end-to-end framework based on Distributionally Robust Learning (DRL), which models differences between source and target domains using a differentiable density ratio estimator. This enables learning calibrated confidence in domain drift scenarios, and the confidence is utilized for Unsupervised Domain Adaptation (UDA) and cross-domain Semi-Supervised Learning (SSL).
Learning Constraint Networks over Unknown Constraint Languages
Christian Bessiere (University of Montpellier), Areski Himeur (University of Montpellier)
OptimizationBenchmark
🎯 What it does: This paper proposes a constraint learning method that does not require prior knowledge of the constraint language, capable of simultaneously inferring appropriate constraint language and constraint networks during the learning process.
Learning Dissemination Strategies for External Sources in Opinion Dynamic Models with Cognitive Biases
Abdullah Al Maruf (University of Washington), Radha Poovendran (University of Washington)
OptimizationGaussian SplattingGraph
🎯 What it does: Propose a multi-channel external information propagation model based on prospect theory, using Gaussian process learning to estimate unknown parameters and designing information propagation strategies through submodular optimization to drive group opinions toward target values.
Learning Efficient Truthful Mechanisms for Trading Networks
Takayuki Osogami (IBM Research Technion Israel Institute Of Technology), Elisheva S. Shamash
OptimizationTabularFinance Related
🎯 What it does: This paper proposes a mechanism learning method for transaction networks, leveraging the Groves mechanism under Bayesian settings to achieve DSIC, efficiency, expected weak budget balance, and expected individual rationality, addressing the infeasibility of simultaneously satisfying these four properties in such networks.
Learning Few-shot Sample-set Operations for Noisy Multi-label Aspect Category Detection
Shiman Zhao (Peking University), Tengjiao Wang (Peking University)
ClassificationTransformerLarge Language ModelTextBenchmark
🎯 What it does: This paper proposes a few-shot multi-label aspect category detection model based on sample set operations (FSO), which can extract discriminative prototypes from a small number of samples and customize query vectors for each category through intersection, difference, and union networks in a noisy multi-label environment;
Learning Gaussian Mixture Representations for Tensor Time Series Forecasting
Jiewen Deng (Southern University of Science and Technology), Xuan Song (Southern University of Science and Technology)
Representation LearningTime Series
🎯 What it does: Proposed a tensor time series prediction framework GMRL based on Gaussian Mixture Models.
Learning Heuristically-Selected and Neurally-Guided Feature for Age Group Recognition Using Unconstrained Smartphone Interaction
Yingmao Miao (Xi'an Jiaotong University), Chao Shen (Xi'an Jiaotong University)
ClassificationRecognitionRecurrent Neural NetworkTime SeriesSequential
🎯 What it does: Constructed the first large-scale dataset for smartphone interactions under unconstrained scenarios, and proposed the AgeCare system to achieve privacy-preserving age group identification and personalized assistance;
Learning in Multi-Memory Games Triggers Complex Dynamics Diverging from Nash Equilibrium
Yuma Fujimoto (Research Center for Integrative Evolutionary Science, SOKENDAI), Kenshi Abe (AI Lab, CyberAgent, Inc)
Reinforcement LearningOrdinary Differential Equation
🎯 What it does: Extend replicator dynamics and gradient ascent algorithm to multi-memory repeated games, prove their continuous-time equivalence, and conduct theoretical analysis and experimental verification on one-memory two-action zero-sum games
Learning Monocular Depth in Dynamic Environment via Context-aware Temporal Attention
Zizhang Wu (ZongmuTech), Jian Pu (Fudan University)
Depth EstimationAutonomous DrivingOptimizationComputational EfficiencyConvolutional Neural NetworkRecurrent Neural NetworkTransformerSupervised Fine-TuningVideo
🎯 What it does: Propose a multi-frame monocular depth estimation network called CTA-Depth, which alternately optimizes depth and camera pose for dynamic scenes by leveraging context-aware temporal attention (CTA) and long-range geometric embedding (LGE).
Learning Object Consistency and Interaction in Image Generation from Scene Graphs
Yangkang Zhang (Zhejiang University), Lingyun Sun (Zhejiang University)
GenerationGraph Neural NetworkTransformerAuto EncoderImageGraph
🎯 What it does: This paper proposes the LOCI method, aiming to solve the problem of objects being overlooked and insufficient interaction between objects in scene graph to image generation through a consistency module and an interaction module.
Learning Preference Models with Sparse Interactions of Criteria
Margot Herin (Sorbonne University), Nataliya Sokolovska (Sorbonne University)
OptimizationComputational EfficiencyRepresentation LearningTabular
🎯 What it does: This paper proposes a learning method based on sparse Mobius representation to learn interaction effects from preference samples in multi-criteria decision making, avoiding the prior k-additive restriction.
Learning Prototype Classifiers for Long-Tailed Recognition
Saurabh Sharma (University of California Santa Barbara), Ambuj Singh (University of California Santa Barbara)
ClassificationImage
🎯 What it does: This paper proposes a classifier based on learnable prototypes, which directly discriminates samples in the representation space using Euclidean distance, thereby eliminating the bias caused by the soft max classifier in long-tailed data due to the relationship between weight norm and sample count.
Learning Small Decision Trees with Large Domain
Eduard Eiben (Royal Holloway University of London), Stefan Szeider (TU Wien)
ClassificationOptimization
🎯 What it does: This study proposes a decision tree learning algorithm under a parameterized complexity framework, proving that the minimum decision tree size (DTS) and minimum depth (DTD) problems are fixed-parameter tractable (FPT) when parameterized only by the solution size and maximum difference (δmax), thereby eliminating dependence on the maximum domain size; simultaneously, it analyzes the approximation error when constructing decision trees using only the minimal support set, providing error bounds and tightness proofs for both binary and non-binary cases.
Learning Summary-Worthy Visual Representation for Abstractive Summarization in Video
Zenan Xu (Sun Yat-sen University), Qun Liu (Huawei Technology)
GenerationKnowledge DistillationRepresentation LearningConvolutional Neural NetworkTransformerVision Language ModelVideoTextMultimodality
🎯 What it does: Propose a framework called Learning Summary-Worthy Visual Representation (SWVR) for the multimodal abstract summarization task.
Learning Survival Distribution with Implicit Survival Function
Yu Ling (Fudan University), Bo Yan (Fudan University)
Biomedical Data
🎯 What it does: Propose an Implicit Survival Function (ISF) that directly estimates the conditional hazard rate using implicit neural representations and time positional encoding, then obtains the survival distribution through numerical integration.
Learning to Act for Perceiving in Partially Unknown Environments
Leonardo Lamanna (Fondazione Bruno Kessler), Paolo Traverso (Fondazione Bruno Kessler)
Robotic IntelligenceConvolutional Neural NetworkImage
🎯 What it does: This paper proposes a method combining online learning with symbolic planning, enabling autonomous agents in partially unknown environments to identify and plan toward situations where target state variables are highly observable, thereby significantly improving perception accuracy.
Learning to Binarize Continuous Features for Neuro-Rule Networks
Wei Zhang (East China Normal University), Jianyong Wang (Tsinghua University)
ClassificationTabular
🎯 What it does: Propose the AutoInt method to automatically and differentiably binarize continuous features in neural rule networks (NRN), and optimize interval boundaries through end-to-end learning.
Learning to Learn from Corrupted Data for Few-Shot Learning
Yuexuan An (Southeast University), Hui Xue (Southeast University)
ClassificationData-Centric LearningMeta LearningConvolutional Neural NetworkRecurrent Neural NetworkTransformerImage
🎯 What it does: Designed a peer collaborative learning framework that uses dual modules to jointly assess sample importance, enhancing the robustness of few-shot learning models against polluted data.
Learning to Self-Reconfigure for Freeform Modular Robots via Altruism Proximal Policy Optimization
Lei Wu (Northwestern Polytechnical University), Zhiwen Yu (Northwestern Polytechnical University)
Robotic IntelligenceMeta LearningReinforcement Learning
🎯 What it does: Designed and implemented a distributed self-reconfiguration algorithm based on multi-agent reinforcement learning, utilizing personalized altruistic factors to enable free-form modular robots to collaboratively avoid conflicts in a continuous action space, achieving efficient self-reconfiguration from one configuration to another.
Learning to Send Reinforcements: Coordinating Multi-Agent Dynamic Police Patrol Dispatching and Rescheduling via Reinforcement Learning
Waldy Joe (Singapore Management University), Hoong Chuin Lau (Singapore Management University)
Reinforcement Learning
🎯 What it does: This paper proposes a multi-agent dynamic police patrol dispatch and rescheduling framework based on reinforcement learning to solve the MADPRP problem.
Learning to Speak from Text: Zero-Shot Multilingual Text-to-Speech with Unsupervised Text Pretraining
Takaaki Saeki (University of Tokyo), Hiroshi Saruwatari (University of Tokyo)
GenerationData SynthesisTransformerGenerative Adversarial NetworkContrastive LearningTextAudio
🎯 What it does: A zero-shot multilingual text-to-speech (TTS) system is studied, which achieves cross-lingual phoneme and prosody transfer through unsupervised multilingual pre-training using only text, enabling speech synthesis for unseen languages.
Learning When to Advise Human Decision Makers
Gali Noti (Harvard University), Yiling Chen (Harvard University)
Recommendation SystemTabular
🎯 What it does: Proposed and experimentally tested a responsive AI-assisted decision system that determines when to provide advice to human decision-makers based on the accuracy of human initial predictions.
Learning When to Use Automatic Tabulation in Constraint Model Reformulation
Carlo Cena (University of Bologna), Felix Ulrich-Oltean (University of York)
ClassificationOptimizationTabular
🎯 What it does: Explored how to use machine learning to predict when to apply automatic tabulation techniques in constraint models to improve the runtime of SAT/CP solvers.
Less Learn Shortcut: Analyzing and Mitigating Learning of Spurious Feature-Label Correlation
Yanrui Du (Harbin Institute of Technology), Bing Qin (Harbin Institute of Technology)
ClassificationExplainability and InterpretabilityData-Centric LearningSupervised Fine-TuningText
🎯 What it does: This paper investigates the 'word-label shortcut' phenomenon in natural language tasks caused by dataset bias in deep learning models, and proposes a task-agnostic training strategy called LessLearn-Shortcut (LLS) to reduce the model's over-reliance on biased words;
Leveraging Argumentation for Generating Robust Sample-based Explanations
Leila Amgoud (CNRS), Henri Trenquier (Toulouse University)
ClassificationExplainability and InterpretabilityTabular
🎯 What it does: Studies how to generate interpretable inductive explanations for black-box classifiers from samples using an argumentation framework.
Levin Tree Search with Context Models
Laurent Orseau (Google DeepMind), Levi H. S. Lelis (University of Alberta)
OptimizationReinforcement LearningMixture of ExpertsBenchmark
🎯 What it does: Propose using a parameterized context model (based on online compression of context trees) instead of neural networks as the strategy for Levin Tree Search (LTS), and prove that the LTS loss function is convex under this model, enabling training of the strategy via convex optimization algorithms. Subsequent experiments were conducted on Sokoban, The Witness, 24-Sliding Tile Puzzle (STP), and Rubik’s Cube.
LGI-GT: Graph Transformers with Local and Global Operators Interleaving
Shuo Yin (Ocean University of China), Guoqiang Zhong (Ocean University of China)
Representation LearningDrug DiscoveryGraph Neural NetworkTransformerGraph
🎯 What it does: Designed an interleaved local-global operation Graph Transformer (LGI-GT) and proposed an attention-based local information enhancement module (EELA).
LGPConv: Learnable Gaussian Perturbation Convolution for Lightweight Pansharpening
Chen-Yu Zhao (University of Electronic Science and Technology of China), Liang-Jian Deng (University of Electronic Science and Technology of China)
Super ResolutionComputational EfficiencyConvolutional Neural NetworkImage
🎯 What it does: Propose a lightweight convolution called LGPConv and build LGPConv-Net based on it for pansharpening high-resolution multispectral images
Lifelong Multi-view Spectral Clustering
Hecheng Cai (Sichuan University), Jiancheng Lv (Sichuan University)
OptimizationRepresentation LearningMeta LearningText
🎯 What it does: Propose a lifelong multi-view spectral clustering framework that achieves cross-task knowledge transfer and update by utilizing an orthogonal basis library and a feature embedding library.
Linear Query Approximation Algorithms for Non-monotone Submodular Maximization under Knapsack Constraint
Canh V. Pham (Phenikaa University), My T. Thai (University of Florida)
OptimizationImageGraph
🎯 What it does: Proposes two constant factor approximation algorithms (DLA and RLA) achieving linear query complexity under the knapsack constraint for non-monotone submodular maximization, where DLA has a deterministic approximation factor of 6+ε and RLA has a randomized approximation factor of 4+ε.
Linguistic More: Taking a Further Step toward Efficient and Accurate Scene Text Recognition
Boqiang Zhang (University of Science and Technology of China), Yongdong Zhang (University of Science and Technology of China)
RecognitionComputational EfficiencyTransformerImage
🎯 What it does: This paper proposes the Linguistic Perception Vision model (LPV), which enhances scene text recognition accuracy by integrating linguistic information into visual models through Cascade Position Attention (CPA) and Global Linguistic Reconstruction Module (GLRM).
LION: Label Disambiguation for Semi-supervised Facial Expression Recognition with Progressive Negative Learning
Zhongjing Du (Sichuan University), Yan Wang (Sichuan University)
RecognitionImage
🎯 What it does: Propose a semi-supervised deep facial expression recognition framework called LION, combining a label disambiguation module and a progressive negative learning module.
LISSNAS: Locality-based Iterative Search Space Shrinkage for Neural Architecture Search
Bhavna Gopal (Duke University), Yiran Chen (Duke University)
Neural Architecture SearchImage
🎯 What it does: Iteratively shrink the search space based on the locality principle to retain a diverse and high-performance subset of networks.
Local and Global: Temporal Question Answering via Information Fusion
Yonghao Liu (Key Laboratory for Symbol Computation and Knowledge Engineering of the Ministry of Education, Jilin University), Renchu Guan (Key Laboratory for Symbol Computation and Knowledge Engineering of the Ministry of Education, Jilin University)
Explainability and InterpretabilityRepresentation LearningGraph Neural NetworkTransformerTextGraph
🎯 What it does: Proposes a temporal knowledge graph question answering model LGQA that integrates local graph structural information with global text semantics.
Local-Global Transformer Enhanced Unfolding Network for Pan-sharpening
Mingsong Li (Shandong University), Gongping Yang (Shandong University)
RestorationSuper ResolutionTransformerImage
🎯 What it does: Proposed a deep unfolding-based local-global Transformer enhanced network (LGTEUN) for sharpening fusion of multispectral images. The network unfolds the iterative process into trainable stages via variational optimization and proximal gradient descent (PGD), with each stage containing a lightweight data module and a powerful local-global Transformer prior module to achieve effective image denoising and information fusion.
Locate, Refine and Restore: A Progressive Enhancement Network for Camouflaged Object Detection
Xiaofei Li (National University of Defense Technology), Dong Chen (National University of Defense Technology)
Object DetectionSegmentationConvolutional Neural NetworkImage
🎯 What it does: Propose an evolutionary enhanced network (PENet) that achieves precise segmentation of camouflaged targets through a three-step process: first locating objects, then refining textures, and finally restoring boundaries.
Low-Confidence Samples Mining for Semi-supervised Object Detection
Guandu Liu (Tsinghua University), Bin Wang (Tsinghua University)
Object DetectionConvolutional Neural NetworkTransformerImage
🎯 What it does: Propose a Low Confidence Sample Mining (LSM) method in semi-supervised object detection, which efficiently mines and learns low-confidence pseudo labels by leveraging an additional pseudo information mining branch and a self-distillation mechanism with low-resolution features.
LSGNN: Towards General Graph Neural Network in Node Classification by Local Similarity
Yuhan Chen (Sun Yat-sen University), Xiaochun Cao (Sun Yat-sen University)
ClassificationGraph Neural NetworkGraphBenchmark
🎯 What it does: Propose a novel graph neural network called LSGNN, which utilizes Local Similarity (LocalSim) to achieve node-level adaptive weighted fusion, and introduces Initial Residual Difference Connection (IRDC) to obtain richer multi-hop information, addressing the performance discrepancy of traditional GNNs on homogeneous and heterogeneous graphs.
MA2CL:Masked Attentive Contrastive Learning for Multi-Agent Reinforcement Learning
Haolin Song (University of Science and Technology of China), Houqiang Li (University of Science and Technology of China)
Representation LearningTransformerReinforcement LearningContrastive LearningImageTabularBenchmark
🎯 What it does: Proposes the MA2CL framework, achieving temporal and team-level representation learning by masking partial agents in multi-agent reinforcement learning and reconstructing their features using attention contrastive learning.
Manifold-Aware Self-Training for Unsupervised Domain Adaptation on Regressing 6D Object Pose
Yichen Zhang (South China University of Technology), Kui Jia (South China University of Technology)
Pose EstimationDomain AdaptationImage
🎯 What it does: Propose a 6D pose estimation method for unsupervised domain adaptation called MAST, which employs self-supervised manifold regularization and self-training strategies to decompose the regression task into coarse classification and fine-grained regression.
MAT: Mixed-Strategy Game of Adversarial Training in Fine-tuning
Zhehua Zhong (Hangzhou Dianzi University), Zhen Wang (Hangzhou Dianzi University)
Representation LearningData-Centric LearningTransformerLarge Language ModelSupervised Fine-TuningText
🎯 What it does: Propose a mixed strategy adversarial training algorithm (MAT), introducing mixed strategy game theory during fine-tuning of pre-trained language models to improve model generalization and robustness.
Matchings under One-Sided Preferences with Soft Quotas
Santhini K. A. (Indian Institute of Technology Madras), Meghana Nasre (Indian Institute of Technology Madras)
OptimizationFlow-based Model
🎯 What it does: Propose introducing soft quotas (each position has upper and lower target values) under a one-sided preference model, and study the multi-objective matching problem between minimizing quota deviation and maximizing ranking/fairness.
Max Markov Chain
Yu Zhang (Arizona State University), Mitchell Bucklew (Arizona State University)
Time SeriesFinance Related
🎯 What it does: This paper proposes Max Markov Chain (MMC), a sequence data modeling method that leverages sparse long-range associations, along with an analytical maximum likelihood parameter estimation and an efficient greedy approximation algorithm.
Maximin-Aware Allocations of Indivisible Chores with Symmetric and Asymmetric Agents
Tianze Wei (City University of Hong Kong), Minming Li (City University of Hong Kong)
Optimization
🎯 What it does: Studied the theoretical existence and algorithmic computation of maximum-minimum perceptual fairness (MMA) allocation and its relaxed forms (MMA1, MMAX) under heterogeneous weights for indivisible tasks.
Mean Payoff Optimization for Systems of Periodic Service and Maintenance
David Klaška (Masaryk University), Vojtěch Řehák (Masaryk University)
OptimizationReinforcement LearningTabular
🎯 What it does: Studied the infinite-horizon cyclic routing problem (IHRRP), proposed random finite memory (RFM) scheduling and periodic scheduling methods, and provided corresponding algorithms.
Measuring a Priori Voting Power in Liquid Democracy
Rachael Colley (Université Toulouse Capitole), Hugo Gilbert (Université Paris Dauphine)
Graph
🎯 What it does: Proposes a prior voting power measure called LD Penrose-Banzhaf in Liquid Democracy, extending the traditional Penrose-Banzhaf index and providing its definition and calculation methods under proxy networks.
Measuring Acoustics with Collaborative Multiple Agents
Yinfeng Yu (Tsinghua University), Fuchun Sun (Tsinghua University)
Robotic IntelligenceConvolutional Neural NetworkRecurrent Neural NetworkReinforcement LearningAgentic AIImageMultimodality
🎯 What it does: This paper studies the task of enabling two robots to collaboratively measure the acoustic characteristics (RIR) of an indoor environment by moving, transmitting/receiving sweep signals.
Measuring and Controlling Divisiveness in Rank Aggregation
Rachael Colley (Université Toulouse Capitole), Carlos Navarrete (Université de Toulouse)
OptimizationTabular
🎯 What it does: Propose a family of parameterized measures of divisiveness (α-divisiveness), study their theoretical properties and relationship with polarization, and provide a polynomial-time method for solving the maximum divisive subgroup problem for Borda scores; simultaneously conduct experimental evaluations on the robustness of the divisiveness measure against missing comparisons and artificially injected rankings, and present a simple injection algorithm (INJECT) to manipulate the divisiveness measure.
Meta-Tsallis-Entropy Minimization: A New Self-Training Approach for Domain Adaptation on Text Classification
Menglong Lu (National University of Defense Technology), Dongsheng Li (National University of Defense Technology)
ClassificationDomain AdaptationGraph Neural NetworkTransformerText
🎯 What it does: Propose Meta-Tsallis-Entropy Minimization (MTEM), a self-training method for text classification domain adaptation, which leverages meta-learning to dynamically learn the Tsallis entropy index for each unlabelled instance, achieving instance-adaptive entropy minimization.
MILD: Modeling the Instance Learning Dynamics for Learning with Noisy Labels
Chuanyang Hu (ShanghaiTech University), Xuming He (ShanghaiTech University)
ClassificationData-Centric LearningContrastive LearningImage
🎯 What it does: Propose a framework for selecting noisy label samples based on instance learning dynamics (MILD), which records the prediction sequence of each sample during training, calculates the difficulty of its memorization and forgetting, combines them into a selection metric CF, automatically determines the threshold using a Weibull Mixture Model, and iteratively selects clean samples and trains the network. The framework can seamlessly integrate with semi-supervised methods such as MixMatch, MixUp, and contrastive learning.
Minimally Supervised Contextual Inference from Human Mobility: An Iterative Collaborative Distillation Framework
Jiayun Zhang (University of California, San Diego), Jingbo Shang (University of California, San Diego)
ClassificationKnowledge DistillationConvolutional Neural NetworkTime Series
🎯 What it does: This paper proposes a minimally supervised mobile context reasoning framework called STCOLAB, which learns contextual information from human mobility data with extremely few labels by utilizing alternating training and collaborative distillation of spatial and temporal modules.
Minimizing Reachability Times on Temporal Graphs via Shifting Labels
Argyrios Deligkas (Royal Holloway, University of London), George Skretas (Hasso Plattner Institute, University of Potsdam)
OptimizationGraph
🎯 What it does: Studied the optimization problem (REACHFAST) of accelerating information propagation by shifting labels in temporal graphs, and provided complexity analysis and algorithm design under various graph structures and constraints.
Mitigating Disparity while Maximizing Reward: Tight Anytime Guarantee for Improving Bandits
Vishakha Patil (Indian Institute of Science), Arindam Khan (Indian Institute of Science)
OptimizationReinforcement LearningTabularBenchmark
🎯 What it does: Proposes an improved multi-armed bandit (IMAB) algorithm under unknown time windows, which ensures optimal cumulative rewards at any moment while allowing each arm to reach its theoretical limit after sufficient opportunities, thus alleviating the gap caused by initial opportunity disparities.
MM-PCQA: Multi-Modal Learning for No-reference Point Cloud Quality Assessment
Zicheng Zhang (Shanghai Jiao Tong University), Guangtao Zhai (Shanghai Jiao Tong University)
Convolutional Neural NetworkGraph Neural NetworkImageMultimodalityPoint Cloud
🎯 What it does: This paper proposes a no-reference multimodal point cloud quality assessment method called MM-PCQA, which combines features from point cloud local sub-models and 2D projections;
MMPN: Multi-supervised Mask Protection Network for Pansharpening
Changjie Chen (Jiangxi University of Finance and Economics), Shengna Wei (Tiangong University)
RestorationSuper ResolutionConvolutional Neural NetworkImage
🎯 What it does: Propose a Multi-Supervised Mask Protection Network (MMPN), which uses a Mask Protection Strategy (MPS) to divide images into two-side masks, separately learning features in different edge regions, and achieving multi-task fusion through a four-branch network, ultimately obtaining high spatial resolution multi-spectral images.
Model Conversion via Differentially Private Data-Free Distillation
Bochao Liu (Chinese Academy of Sciences), Shiming Ge (Chinese Academy of Sciences)
GenerationData SynthesisSafty and PrivacyKnowledge DistillationGenerative Adversarial NetworkImage
🎯 What it does: Convert a pre-trained teacher model into a student model with differential privacy protection through data-agnostic distillation.
Model Predictive Control with Reach-avoid Analysis
Dejin Ren (State Key Laboratory of Computer Science Institute of Software Chinese Academy of Sciences), Bai Xue (State Key Laboratory of Computer Science Institute of Software Chinese Academy of Sciences)
Optimization
🎯 What it does: A learning-based model predictive control (RAMPC) algorithm based on reach-avoid analysis is designed to solve constrained reach-avoid optimization problems.
Modeling Moral Choices in Social Dilemmas with Multi-Agent Reinforcement Learning
Elizaveta Tennant (University College London), Mirco Musolesi (University College London)
OptimizationReinforcement LearningTabular
🎯 What it does: In three classic social dilemma games (Iterated Prisoner's Dilemma, Volunteer's Dilemma, Stag Hunt), Q-learning is used to learn moral reward functions based on different ethical theories (utilitarianism, deontology, virtue equality, virtue benevolence, hybrid virtue). The behaviors and social outcomes of these moral agents, as well as selfish agents and traditional game theory agents, are systematically evaluated in two-player interactions.
Modeling with Homophily Driven Heterogeneous Data in Gossip Learning
Abhirup Ghosh (University of Cambridge), Cecilia Mascolo (University of Cambridge)
OptimizationFederated LearningImageGraphTabular
🎯 What it does: Propose a neighbor-weighted aggregation strategy based on softmax distribution in the Gossip Learning environment to enhance convergence speed under heterogeneous data scenarios.
MolHF: A Hierarchical Normalizing Flow for Molecular Graph Generation
Yiheng Zhu (Zhejiang University), Jian Wu (Zhejiang University)
Drug DiscoveryGraph Neural NetworkFlow-based ModelGraphBiomedical Data
🎯 What it does: Propose a hierarchical normalization flow model called MolHF that generates molecular graphs using a coarse-to-fine approach.
Moral Planning Agents with LTL Values
Umberto Grandi (University of Toulouse), Timothy Parker (University of Toulouse)
🎯 What it does: Proposed the Moral Planning Agent (MPA) framework, which uses Linear Temporal Logic (LTL) to represent priority value bases for evaluating and comparing joint plans and individual plans.
More for Less: Safe Policy Improvement with Stronger Performance Guarantees
Patrick Wienhöft, Nils Jansen (Radboud University)
Reinforcement LearningTabularBenchmark
🎯 What it does: Under the Safe Policy Improvement (SPI) framework of offline reinforcement learning, a novel MDP transformation method is proposed, converting any finite MDP into a two-successor MDP (2sMDP), along with corresponding dataset transformation, achieving tighter performance improvement bounds and lower sample complexity.
Multi-Agent Intention Recognition and Progression
Michael Dann (RMIT University), John Thangarajah (RMIT University)
OptimizationReinforcement Learning
🎯 What it does: This paper proposes a multi-agent intent recognition and scheduling framework named I GR, which utilizes online goal recognition (based on KL divergence in reinforcement learning) to infer the current goals of other agents in the same scenario, and embeds the inferred results into an MCTS intent scheduler to achieve prediction of other agents' behaviors and optimization of self-actions.
Multi-Agent Systems with Quantitative Satisficing Goals
Senthil Rajasekaran (Rice University), Moshe Y. Vardi (Rice University)
Computational Efficiency
🎯 What it does: This paper proposes the use of multi-threshold satisficing goals in concurrent discounted sum games to study pure strategy Nash equilibria;
Multi-level Graph Contrastive Prototypical Clustering
Yuchao Zhang (Northwestern Polytechnical University), Qi Wang (Northwestern Polytechnical University)
Representation LearningGraph Neural NetworkContrastive LearningGraphBenchmark
🎯 What it does: Propose an end-to-end multi-layer graph contrastive prototype clustering framework called MLG-CPC, which can simultaneously learn representations at different granularities and perform unsupervised graph clustering.
Multi-Modality Deep Network for JPEG Artifacts Reduction
Xuhao Jiang (Fudan University), Liquan Shen (Shanghai University)
RestorationConvolutional Neural NetworkVision Language ModelGenerative Adversarial NetworkContrastive LearningImageTextMultimodality
🎯 What it does: Propose a text-guided generative adversarial network (TGJAR) that enhances JPEG image restoration at extremely low bit rates through image-text fusion (global and local) and contrastive learning.
Multi-objective Optimization-based Selection for Quality-Diversity by Non-surrounded-dominated Sorting
Ren-Jian Wang (Nanjing University), Qiang Fu (Tencent)
OptimizationReinforcement LearningBenchmark
🎯 What it does: Proposed a new parent selection method called Non-Encircled Dominance Sorting (NSS) and integrated it into QD algorithms such as MAP-Elites; experimental results verified its superiority on multiple benchmark tasks.
Multi-Robot Coordination and Layout Design for Automated Warehousing
Yulun Zhang (Carnegie Mellon University), Jiaoyang Li (Carnegie Mellon University)
OptimizationRobotic IntelligenceConvolutional Neural Network
🎯 What it does: For multi-robot path planning in automated warehouses, optimizing the warehouse layout to enhance throughput and scalability.
Multi-Scale Subgraph Contrastive Learning
Yanbei Liu (Tiangong University), Zhitao Xiao (Beihang University)
ClassificationRepresentation LearningGraph Neural NetworkContrastive LearningGraph
🎯 What it does: Proposed a Multi-Scale Subgraph Contrastive Learning (MSSGCL) framework that generates global and local views through subgraph sampling, constructing three types of contrastive relationships (global-global, global-local, and local-local) to enhance graph-level representation learning.
Multi-Task Learning via Time-Aware Neural ODE
Feiyang Ye (Southern University of Science and Technology), Ivor W. Tsang (University of Technology Sydney)
ImageBenchmarkOrdinary Differential Equation
🎯 What it does: Proposed the NORMAL method, which achieves task-specific feature transformation in multi-task learning by learning task positions (time-aware task positions) within Neural ODE.
Multi-view Contrastive Learning Hypergraph Neural Network for Drug-Microbe-Disease Association Prediction
Luotao Liu (Huazhong Agricultural University), Wen Zhang (Huazhong Agricultural University)
Drug DiscoveryGraph Neural NetworkContrastive LearningBiomedical Data
🎯 What it does: This paper constructs a hypergraph of drug-microbe-disease tripartite associations and employs a multi-view contrastive learning hypergraph neural network (MCHNN) for association prediction.
Multi-View Robust Graph Representation Learning for Graph Classification
Guanghui Ma (Beihang University), Hong Zhang (National Computer Network Emergency Response Technical Team Coordination Center of China)
ClassificationGraph Neural NetworkContrastive LearningGraphBenchmark
🎯 What it does: Propose a multi-view graph representation learning framework MGRL, enhancing the robustness of graph classification through instance view consistency and class view discriminative learning.
MultiPar-T: Multiparty-Transformer for Capturing Contingent Behaviors in Group Conversations
Dong Won Lee (Massachusetts Institute of Technology), Hae Won Park (Massachusetts Institute of Technology)
ClassificationRecurrent Neural NetworkTransformerMultimodality
🎯 What it does: Studied a multi-participant Transformer called MultiPar-T for capturing follower behaviors in group conversations, and achieved interactive modeling for multiple participants.
Musical Voice Separation as Link Prediction: Modeling a Musical Perception Task as a Multi-Trajectory Tracking Problem
Emmanouil Karystinaios (Johannes Kepler University Linz), Gerhard Widmer (Johannes Kepler University Linz)
Convolutional Neural NetworkRecurrent Neural NetworkGraph Neural NetworkText
🎯 What it does: This paper models the sound separation problem as a multi-target tracking (MTT) problem, using graph neural networks to embed each note and determining adjacent notes within the same voice through link prediction to recover complete voice trajectories.
Negative Flux Aggregation to Estimate Feature Attributions
Xin Li (Wayne State University), Dongxiao Zhu (Wayne State University)
Explainability and InterpretabilityConvolutional Neural NetworkImage
🎯 What it does: Proposes the Negative Flux Aggregation (NeFLAG) method based on vector divergence and flux for DNN explanation without benchmarks or path integrals;
NerCo: A Contrastive Learning Based Two-Stage Chinese NER Method
Zai Zhang (Xi'an Jiaotong University), Qinghua Zheng (Xi'an Jiaotong University)
RecognitionRepresentation LearningTransformerSupervised Fine-TuningContrastive LearningText
🎯 What it does: This paper proposes a two-stage Chinese Named Entity Recognition framework called NerCo, first aggregating representations of words with the same entity type through contrastive learning, and then performing fine-tuning based on traditional sequence labeling.
NeuPSL: Neural Probabilistic Soft Logic
Connor Pryor (University of California Santa Barbara), Lise Getoor (University of California Santa Barbara)
ClassificationOptimizationComputational EfficiencyImageGraph
🎯 What it does: Propose the NeuPSL neuro-symbolic framework, integrating outputs of deep neural networks with probabilistic soft logic (PSL) to achieve differentiable joint reasoning and learning.