IJCAI 2023 Papers with AI Summaries
International Joint Conference on Artificial Intelligence · 639 papers
→ IJCAI 2023 papers with code (241)
Each paper below shows an AI-generated one-line summary. Get the full 6-part summary (innovation, method, data, results, limitations) and search all 639 IJCAI 2023 papers by keyword, author or institution —
free trial on arXivSub.
3D Surface Super-resolution from Enhanced 2D Normal Images: A Multimodal-driven Variational AutoEncoder Approach
Wuyuan Xie (Shenzhen University), Miaohui Wang (Shenzhen University)
Super ResolutionAuto EncoderImageMultimodality
🎯 What it does: Propose a multi-modal driven variational autoencoder (mmVAE) framework that utilizes RGB and normal maps to achieve 3D surface super-resolution, restoring high-resolution surface details.
A Bitwise GAC Algorithm for Alldifferent Constraints
Zhe Li (National University of Defense Technology), Zhanshan Li (Jilin University)
OptimizationComputational EfficiencyBenchmark
🎯 What it does: Proposed a GAC algorithm called Alldiff bit based on bitwise operations for efficiently handling all-different constraints.
A Canonicalization-Enhanced Known Fact-Aware Framework For Open Knowledge Graph Link Prediction
Yilin Wang (National University of Defense Technology), Xicheng Lu (National University of Defense Technology)
Representation LearningGraph Neural NetworkTransformerGraphRetrieval-Augmented Generation
🎯 What it does: Propose a two-stage framework named CEKFA to enhance link prediction performance in open knowledge graphs (OpenKG) through the normalization of relation phrases (RP) and triplets.
A Comparative Study of Ranking Formulas Based on Consistency
Badran Raddaoui (SAMOVAR, T'el'ecom SudParis, Institut Polytechnique de Paris), Said Jabbour (CRIL - CNRS, Université d'Artois)
🎯 What it does: Proposed a formula ranking framework based on the consistency structure of a knowledge base, and systematically compared multiple ranking semantics.
A Diffusion Model with Contrastive Learning for ICU False Arrhythmia Alarm Reduction
Feng Wu (Xi'an Jiaotong University), Li-wei H. Lehman (Massachusetts Institute of Technology)
Anomaly DetectionTransformerDiffusion modelContrastive LearningTime SeriesBiomedical DataElectronic Health RecordsElectrocardiogram
🎯 What it does: Propose a generative method based on conditional diffusion models and contrastive learning for detecting false alarms in ICU arrhythmia alerts.
A Dual Semantic-Aware Recurrent Global-Adaptive Network for Vision-and-Language Navigation
Liuyi Wang (Tongji University), Qijun Chen (Tongji University)
Recurrent Neural NetworkTransformerLarge Language ModelVision Language ModelVision-Language-Action ModelImageTextMultimodality
🎯 What it does: Propose a Dual Semantic Perception Recursive Global Adaptive Network (DSRG) to enhance guided semantics in visual and language inputs and improve memory mechanisms to boost visual language navigation performance.
A Fast Adaptive Randomized PCA Algorithm
Xu Feng (Tsinghua University), Wenjian Yu (Tsinghua University)
OptimizationComputational EfficiencyRepresentation LearningImageGraphTabular
🎯 What it does: Proposes a fast adaptive randomized PCA algorithm called farPCA, which can automatically determine the rank of low-rank decomposition under a given error tolerance.
A Fast Algorithm for Consistency Checking Partially Ordered Time
Leif Eriksson (Linkoping University), Victor Lagerkvist (Linkoping University)
OptimizationComputational Efficiency
🎯 What it does: This paper proposes a new algorithm for consistency checking in partially ordered time (POT), reducing its single-exponential upper bound from 0.368^n n to 0.2601^n n.
A Fast Maximum k-Plex Algorithm Parameterized by the Degeneracy Gap
Zhengren Wang (University of Electronic Science and Technology of China), Mingyu Xiao (University of Electronic Science and Technology of China)
OptimizationGraph
🎯 What it does: Proposed a fast maximum k-plex algorithm based on the degradation gap, aiming to find the largest k-plex from a given graph.
A Generalized Deep Markov Random Fields Framework for Fake News Detection
Yiqi Dong (Tianjin University), Di Jin (Tianjin University)
ClassificationGraph Neural NetworkTransformerMultimodality
🎯 What it does: Propose a semi-supervised fake news detection framework based on deep Markov random fields (GDMRFF) and implement its specific model ALGM, which can integrate multi-modal information of text and images and eliminate event interference to improve the effectiveness of fake news identification.
A Hierarchical Approach to Population Training for Human-AI Collaboration
Yi Loo (Singapore University of Technology and Design), Malika Meghjani (Singapore University of Technology and Design)
Convolutional Neural NetworkRecurrent Neural NetworkReinforcement Learning
🎯 What it does: Propose an algorithm based on hierarchical population training (HiPT), enabling deep reinforcement learning agents to learn multiple optimal response strategies and dynamically switch between them when facing diverse partners, thereby enhancing collaboration with unknown partners.
A Large-Scale Film Style Dataset for Learning Multi-frequency Driven Film Enhancement
Zinuo Li (Shenzhen Institute of Advanced Technology, Chinese Academy of Sciences), Chi-Man Pun (Shenzhen Institute of Advanced Technology, Chinese Academy of Sciences)
Image TranslationConvolutional Neural NetworkImage
🎯 What it does: Proposed the FilmSet large-scale movie-style image dataset and developed the FilmNet framework based on the Laplacian pyramid, achieving multi-band image style transfer;
A Logic-based Approach to Contrastive Explainability for Neurosymbolic Visual Question Answering
Thomas Eiter (TU Wien), Johannes Oetsch (TU Wien)
Object DetectionExplainability and InterpretabilityConvolutional Neural NetworkRecurrent Neural NetworkVision Language ModelContrastive LearningImageTextBenchmark
🎯 What it does: Proposed a logic-based contrastive explanation framework that generates contrastive explanations (CE) using a neuro-symbolic VQA architecture (perception module + LSTM + ASP reasoning)
A Low Latency Adaptive Coding Spike Framework for Deep Reinforcement Learning
Lang Qin (Zhejiang University), Huajin Tang (Zhejiang University)
Computational EfficiencySpiking Neural NetworkReinforcement LearningVideo
🎯 What it does: Proposed a low-latency adaptive coding spiking framework (ACSF) that encodes states and decodes value functions through learnable matrix multiplication, enabling direct training that scales to both online and offline reinforcement learning;
A Mathematical Runtime Analysis of the Non-dominated Sorting Genetic Algorithm III (NSGA-III)
Simon Wietheger (University of Potsdam), Benjamin Doerr (Institut Polytechnique de Paris)
OptimizationBenchmark
🎯 What it does: This paper conducts the first analysis of the mathematical runtime of NSGA-III on the three-objective ONEMINMAX (3-OMM) benchmark, proving that the algorithm does not lose Pareto optimal solutions and can cover the complete Pareto front within an expected O(n log n) iterations.
A Multi-Modal Neural Geometric Solver with Textual Clauses Parsed from Diagram
Ming-Liang Zhang (Chinese Academy of Sciences), Cheng-Lin Liu (Chinese Academy of Sciences)
Explainability and InterpretabilityConvolutional Neural NetworkRecurrent Neural NetworkTransformerLarge Language ModelVision Language ModelMultimodality
🎯 What it does: Propose a multi-modal neural geometric solver, PGPSNet, which integrates text clauses (structural clauses and semantic clauses) parsed from graphics with images and text questions to generate interpretable solution programs based on geometric theorems, and constructs a large-scale annotated dataset called PGPS9K.
A New ANN-SNN Conversion Method with High Accuracy, Low Latency and Good Robustness
Bingsen Wang (Peking University), Yuan Wang (Peking University)
ClassificationComputational EfficiencyKnowledge DistillationConvolutional Neural NetworkSpiking Neural NetworkImage
🎯 What it does: Proposed a new conversion method between ANN and SNN, utilizing the StepReLU activation function and leakage mechanism to achieve high-precision SNN with a single time step.
A New Variable Ordering for In-processing Bounded Variable Elimination in SAT Solvers
Shuolin Li (Aix Marseille Univ), Felip Manyà (Artificial Intelligence Research Institute)
OptimizationBenchmark
🎯 What it does: This paper proposes a new variable ordering method called ESA (Elimination by Activity Scheduling), which is integrated into several mainstream CDCL SAT solvers to improve the bounded variable elimination (BVE) process during solving.
A Noisy-Label-Learning Formulation for Immune Repertoire Classification and Disease-Associated Immune Receptor Sequence Identification
Mingcai Chen (Nanjing University), Jianhua Yao (Tencent AI Lab)
ClassificationTransformerSupervised Fine-TuningBiomedical Data
🎯 What it does: The study transforms the immune receptor classification problem into a sequence-level learning task with noisy labels, achieving sequence-level and bag-level classification through robust training.
A Novel Demand Response Model and Method for Peak Reduction in Smart Grids -- PowerTAC
Sanjay Chandlekar (International Institute of Information Technology), Sujit Gujar (International Institute of Information Technology)
OptimizationTabularTime Series
🎯 What it does: Proposed a demand response model that models the load reduction probability as an exponential function and introduces a load reduction rate parameter, and developed an optimal budget allocation algorithm MJS-EXPRESPONSE and an online learning algorithm MJSUCB-EXPRESPONSE in PowerTAC simulations to reduce peak load.
A Novel Learnable Interpolation Approach for Scale-Arbitrary Image Super-Resolution
Jiahao Chao (East China Normal University), Zhengfeng Yang (East China Normal University)
Super ResolutionMeta LearningConvolutional Neural NetworkImage
🎯 What it does: Proposed a learnable interpolation module and a scale-aware channel attention module to achieve single-image super-resolution at arbitrary scales.
A Refined Upper Bound and Inprocessing for the Maximum K-plex Problem
Hua Jiang (Yunnan University), Wei Zhou (Yunnan University)
OptimizationComputational EfficiencyGraph
🎯 What it does: This paper proposes a new branch-and-bound algorithm (DiseMKP), which introduces an improved upper bound (DisePUB) and incremental pruning (inprocessing) strategies into solving the maximum k-plex problem (MKP). These strategies dynamically prune the graph during the search process, significantly reducing the search space.
A Regular Matching Constraint for String Variables
Roberto Amadini (University of Bologna), Peter J. Stuckey (Monash University)
OptimizationComputational EfficiencyText
🎯 What it does: Proposed and implemented a specialized propagator for match constraints on string variables, capable of returning the leftmost match position of a regular pattern within a string.
A Rigorous Risk-aware Linear Approach to Extended Markov Ratio Decision Processes with Embedded Learning
Alexander Zadorojniy (IBM Research), Orit Davidovich (IBM Research)
OptimizationReinforcement LearningBenchmark
🎯 What it does: Proposed a risk-aware extended Markov ratio decision process (EMRDP) and provided a strongly polynomial solving algorithm under the monotonicity assumption.
A Rule-Based Modal View of Causal Reasoning
Emiliano Lorini (IRIT)
Explainability and InterpretabilityComputational Efficiency
🎯 What it does: This paper proposes a rule-based causal inference semantics and constructs multiple modal languages on it, supporting the expression and reasoning of causal necessity, possibility, interventionist conditions, and Lewis-style conditions.
A Solution to Co-occurence Bias: Attributes Disentanglement via Mutual Information Minimization for Pedestrian Attribute Recognition
Yibo Zhou (Beihang University), Yuran Cao (Beihang University)
RecognitionRepresentation LearningConvolutional Neural NetworkContrastive LearningImage
🎯 What it does: Propose an attribute decoupling feature learning method that utilizes mutual information minimization to alleviate attribute co-occurrence bias in pedestrian attribute recognition.
A Symbolic Approach to Computing Disjunctive Association Rules from Data
Said Jabbour (CRIL, Universite d'Artois & CNRS), Lakhdar Sais (CRIL, Universite d'Artois & CNRS)
Tabular
🎯 What it does: This paper proposes a symbolization method based on k-disjunctive support for mining disjunctive association rules from transaction databases containing missing items;
A Unification Framework for Euclidean and Hyperbolic Graph Neural Networks
Mehrdad Khatir (Virginia Tech), Chandan K. Reddy (Virginia Tech)
ClassificationComputational EfficiencyGraph Neural NetworkGraph
🎯 What it does: This paper proposes a pseudo-Poincaré framework that converts traditional Euclidean GNNs into hyperbolic geometric networks capable of capturing hierarchical structures through a single hyperplane regularization layer and a Riemannian geometry optimizer.
A Unifying Formal Approach to Importance Values in Boolean Functions
Hans Harder (University of Paderborn), Clemens Dubslaff (Eindhoven University of Technology)
Explainability and InterpretabilityComputational EfficiencyBenchmark
🎯 What it does: Proposes a unified IVF framework, formally defining the importance of variables in Boolean functions, and provides multiple instances and symbolic computation methods.
Abstraction of Nondeterministic Situation Calculus Action Theories
Bita Banihashemi (Ronin Institute), Yves Lesperance (York University)
🎯 What it does: Proposes an abstract framework for non-deterministic situation calculus action theory, supporting strategy synthesis and execution monitoring between high-level abstractions and low-level specifics.
Accurate MRI Reconstruction via Multi-Domain Recurrent Networks
Jinbao Wei (University of Science and Technology of China), Xun Chen (University of Science and Technology of China)
RestorationConvolutional Neural NetworkRecurrent Neural NetworkBiomedical DataMagnetic Resonance Imaging
🎯 What it does: Developed a multi-domain recursive network (MDR-Net), which utilizes multi-domain learning blocks to interactively learn in the spatial domain and frequency domain, achieving progressive reconstruction of under-sampled MRI data.
Acoustic NLOS Imaging with Cross Modal Knowledge Distillation
Ui-Hyeon Shin (Sungkyunkwan University), Kwangsu Kim (Sungkyunkwan University)
RestorationKnowledge DistillationConvolutional Neural NetworkGenerative Adversarial NetworkMultimodalityPhysics RelatedAudio
🎯 What it does: Propose a cross-modal knowledge distillation (CMKD) framework, transferring knowledge from a trained image-to-depth network to an audio-to-depth network for acoustic non-line-of-sight (NLOS) imaging, enabling scene reconstruction using only acoustic signals.
Action Recognition with Multi-stream Motion Modeling and Mutual Information Maximization
Yuheng Yang (Jilin University), Kui Ren (Zhejiang University)
RecognitionGraph Neural NetworkVideo
🎯 What it does: This paper proposes a multi-stream action recognition framework that integrates low-level (coordinates, bone length, velocity, etc.) and high-level (angular acceleration) motion features, and supervises the features through mutual information maximization.
Action Space Reduction for Planning Domains
Harsha Kokel (IBM T.J. Watson Research Center), Shirin Sohrabi (IBM T.J. Watson Research Center)
OptimizationReinforcement Learning
🎯 What it does: Automate the reduction of action labels in classical planning tasks, introduce the concept of actionable mutex groups, and achieve label space compression through parameter seed set problems (solved using delete-free planning).
Active Visual Exploration Based on Attention-Map Entropy
Adam Pardyl (IDEAS NCBR), Tomasz Trzciński (IDEAS NCBR)
ClassificationRestorationSegmentationTransformerAuto EncoderImage
🎯 What it does: This paper proposes an active visual exploration method based on Transformer attention entropy (Attention-Map Entropy, AME), which directly selects the next frame perspective (glimpse) by leveraging the attention uncertainty within the MAE autoencoder, without requiring additional sampling decision heads or auxiliary losses;
Actor-Multi-Scale Context Bidirectional Higher Order Interactive Relation Network for Spatial-Temporal Action Localization
Jun Yu (University of Science and Technology of China), Jinze Wu (iFLYTEK Co., Ltd.)
Object DetectionConvolutional Neural NetworkTransformerVideo
🎯 What it does: Proposed AMCRNet, which improves video action detection through multi-scale context and bidirectional high-order interaction relationships;
ActUp: Analyzing and Consolidating tSNE and UMAP
Andrew Draganov (Aarhus University), Cigdem Aslay (Aarhus University)
OptimizationComputational EfficiencyRepresentation LearningImageBiomedical Data
🎯 What it does: This paper conducts a unified analysis of tSNE and UMAP, and proposes the Gradient Dimensionality Reduction (GDR) algorithm, which can simultaneously achieve embeddings of both methods.
Adaptive Estimation Q-learning with Uncertainty and Familiarity
Xiaoyu Gong (Jilin University), Zongze Li (Jilin University)
Reinforcement Learning
🎯 What it does: Propose the Adaptive Estimation Q-learning (AEQ) method, which dynamically adjusts Q-value estimation by combining uncertainty and familiarity, thereby mitigating overestimation and underestimation biases in offline reinforcement learning.
Adaptive Path-Memory Network for Temporal Knowledge Graph Reasoning
Hao Dong (Chinese Academy of Sciences), Yanjie Fu (University of Central Florida)
Representation LearningGraph Neural NetworkGraphTime Series
🎯 What it does: Propose a relation-based adaptable path memory network (DaeMon), which performs knowledge graph reasoning at future time points by leveraging relational chain (path) information between query entities and candidates in historical time.
Adaptive Reward Shifting Based on Behavior Proximity for Offline Reinforcement Learning
Zhe Zhang (Nanjing University of Aeronautics and Astronautics), Xiaoyang Tan (Nanjing University of Aeronautics and Astronautics)
Reinforcement LearningBenchmark
🎯 What it does: Propose a bidirectional reward shifting (PBRS) method based on adaptive behavior similarity, which uses positive and negative reward shifts to respectively control exploration of actions in discrete distribution and external distribution, thereby addressing the distribution shift problem in offline RL.
Adaptive Sparse ViT: Towards Learnable Adaptive Token Pruning by Fully Exploiting Self-Attention
Xiangcheng Liu (Peking University), Guodong Guo (Baidu Research)
ClassificationComputational EfficiencyKnowledge DistillationTransformerImage
🎯 What it does: Propose an adaptive sparse ViT (AS-ViT), which computes token importance by leveraging intermediate results of multi-head self-attention (MHSA) and achieves sample-adaptive token pruning without additional computation using learnable thresholds.
Advancing Post-Hoc Case-Based Explanation with Feature Highlighting
Eoin M. Kenny (Massachusetts Institute of Technology), Mark T. Keane (University College Dublin)
ClassificationExplainability and InterpretabilityConvolutional Neural NetworkImage
🎯 What it does: Propose two post-hoc case-based reasoning (CBR) methods combined with feature highlighting techniques, which can extract and display multiple salient features from image classification models for each prediction, and associate these features with corresponding cases in the training data to generate interpretable prediction explanations.
Adversarial Amendment is the Only Force Capable of Transforming an Enemy into a Friend
Chong Yu (Fudan University), Zhongxue Gan (Fudan University)
ClassificationObject DetectionSegmentationAdversarial AttackConvolutional Neural NetworkImage
🎯 What it does: This study proposes the AdvAmd method, which transforms adversarial examples into a 'friendly force' that improves model accuracy on clean data through fine-grained attacks, generating intermediate samples, incorporating auxiliary batch normalization (BN), and using adaptive loss, while maintaining adversarial robustness.
Adversarial Behavior Exclusion for Safe Reinforcement Learning
Md Asifur Rahman (Wake Forest University), Sarra Alqahtani (Wake Forest University)
Explainability and InterpretabilityAdversarial AttackReinforcement LearningSequentialBenchmark
🎯 What it does: Proposed a task-agnostic safety protection framework called AdvEx-RL, which trains an adversary policy to identify and eliminate all actions violating safety constraints, then learns a safety policy that acts as a safety valve to block dangerous actions during runtime.
Adversarial Contention Resolution Games
Giorgos Chionas (University of Liverpool), Piotr Krysta (University of Liverpool)
Optimization
🎯 What it does: The study addresses conflict resolution games involving selfish players on shared channels, proposing and constructing a deterministic algorithm termed 'Adversarial Equilibrium' (AE), and analyzing its performance and price of stability/price of anarchy (PoS/PoA).
Algorithmics of Egalitarian versus Equitable Sequences of Committees
Eva Michelle Deltl (Technische Universitaet Berlin), Robert Bredereck (Technische Universitaet Clausthal)
Optimization
🎯 What it does: This paper studies the algorithms and complexity of selecting multi-layered committee sequences, focusing on the problem of selecting a committee of size k at each time point, ensuring that at least x voters are satisfied at each time point, and each voter is satisfied at least y times across a total of τ time points (equilibrium or fairness);
Align, Perturb and Decouple: Toward Better Leverage of Difference Information for RSI Change Detection
Supeng Wang (Fudan University), Wenbing Zhu (Rongcheer)
SegmentationConvolutional Neural NetworkGraph Neural NetworkImage
🎯 What it does: This paper proposes the APD framework, which enhances RSI change detection through alignment, perturbation, and decoupling operations.
ALL-E: Aesthetics-guided Low-light Image Enhancement
Ling Li (Nanjing University of Aeronautics and Astronautics), Songcan Chen (Nanjing University of Aeronautics and Astronautics)
RestorationReinforcement LearningImage
🎯 What it does: Propose a low-light image enhancement framework ALL-E that utilizes aesthetic assessment as a reward, employing reinforcement learning to achieve pixel-level recursive enhancement
An Effective and Efficient Time-aware Entity Alignment Framework via Two-aspect Three-view Label Propagation
Li Cai (East China Normal University), Man Lan (East China Normal University)
Computational EfficiencyRepresentation LearningGraphTime Series
🎯 What it does: Propose a non-neural network time-aware entity alignment framework called LightTEA, which achieves efficient alignment of temporal knowledge graphs using label propagation, sparse similarity, Sinkhorn operations, and iterative learning.
An Empirical Study on the Language Modal in Visual Question Answering
Daowan Peng (Huazhong University of Science and Technology), Dangyang Chen (Ping An Property & Casualty Insurance Company of China Ltd)
TransformerVision Language ModelContrastive LearningImageTextMultimodality
🎯 What it does: This paper investigates the impact of the language modality on model bias in visual question answering tasks through empirical experiments, verifying the dominant role of the suffix (visual concept) in linguistic bias, and enhancing out-of-distribution (OOD) performance by training models with variant questions. Subsequently, an adversarial contrastive learning and feature fusion method for debiasing is proposed, which improves the accuracy of multiple baseline models on VQA-CPv2.
An Ensemble Approach for Automated Theorem Proving Based on Efficient Name Invariant Graph Neural Representations
Achille Fokoue (IBM Research), Radu Marinescu (IBM Research)
Computational EfficiencyRepresentation LearningGraph Neural NetworkReinforcement LearningGraph
🎯 What it does: This paper studies a nominal-invariant graph neural network representation and ensemble strategy for automated theorem proving, achieving higher solving rates on multiple domain datasets.
An Exact Algorithm for the Minimum Dominating Set Problem
Hua Jiang (Yunnan University), Zhifei Zheng (Yunnan University)
OptimizationComputational EfficiencyGraph
🎯 What it does: Designed a lower bound based on 2-hop adjacency, and proposed the EMOS exact branch-and-bound algorithm to solve the minimum dominating set problem.
An Experimental Comparison of Multiwinner Voting Rules on Approval Elections
Piotr Faliszewski (AGH University), Stanisław Szufa (AGH University)
OptimizationTabular
🎯 What it does: This paper proposes a method to compare multi-member committee voting rules based on the similarity of committees generated in elections, defining distances between candidates and extending them to committee distances;
Analyzing and Combating Attribute Bias for Face Restoration
Zelin Li (Southern University of Science and Technology), Bo Tang (Southern University of Science and Technology)
RestorationSuper ResolutionConvolutional Neural NetworkVision Language ModelAuto EncoderGenerative Adversarial NetworkImage
🎯 What it does: This paper proposes the DebiasFR framework, aiming to address the attribute bias problem that occurs during facial restoration, i.e., the restored facial attributes (such as gender and age) significantly differ from those in the original low-resolution image or real high-resolution image.
Analyzing Intentional Behavior in Autonomous Agents under Uncertainty
Filip Cano Córdoba, Bettina Könighofer
Autonomous DrivingReinforcement LearningTabular
🎯 What it does: Propose a quantification method based on Markov Decision Processes (MDP), using a probabilistic model to check evidence of autonomous agents achieving intentional events in uncertain environments
Annealing Genetic-based Preposition Substitution for Text Rubbish Example Generation
Chen Li (China University of Petroleum (East China)), Honglong Chen (China University of Petroleum (East China))
Adversarial AttackText
🎯 What it does: This paper proposes an Annealing Genetic-based preposition replacement algorithm (AGPS) for generating text spam samples;
Answer Mining from a Pool of Images: Towards Retrieval-Based Visual Question Answering
Abhirama Subramanyam Penamakuri (Indian Institute of Technology Jodhpur), Anand Mishra (Indian Institute of Technology Jodhpur)
GenerationRetrievalConvolutional Neural NetworkTransformerVision Language ModelImageMultimodalityBenchmarkRetrieval-Augmented Generation
🎯 What it does: Proposes the Retrieval-based Visual Question Answering (RETVQA) task, which involves retrieving relevant images from a large set of images given a question, and then generating a natural language answer.
Anticipatory Fictitious Play
Alex Cloud (Riot Games), Wesley Kerr (Riot Games)
Reinforcement Learning
🎯 What it does: Proposed and analyzed the Anticipatory Fictitious Play (AFP) algorithm, improving upon traditional Fictitious Play to achieve faster convergence and superior empirical performance.
Appearance Prompt Vision Transformer for Connectome Reconstruction
Rui Sun (University of Science and Technology of China), Feng Wu (University of Science and Technology of China)
SegmentationTransformerPrompt EngineeringContrastive LearningImageBiomedical Data
🎯 What it does: Proposes an APViT framework based on Vision Transformer for neural network reconstruction of 3D electron microscopy images, combining affinity learning with metric learning.
Approximate Envy-Freeness in Graphical Cake Cutting
Sheung Man Yuen (National University of Singapore), Warut Suksompong (National University of Singapore)
OptimizationGraph
🎯 What it does: Studies approximate envy-freeness in connected allocations for graph cake cutting, and provides algorithms for various graph types and homogeneous/heterogeneous valuation scenarios.
Approximate Inference in Logical Credal Networks
Radu Marinescu (IBM Research), Ryan Riegel (IBM Research)
OptimizationComputational EfficiencyGraphBiomedical Data
🎯 What it does: Proposed an approximate inference algorithm called ARIEL for Logical Credibility Networks (LCN), which estimates the posterior probability intervals of atomic propositions by iteratively passing upper and lower bound messages on factor graphs.
Approximating Fair Division on D-Claw-Free Graphs
Zbigniew Lonc (Warsaw University of Technology)
Graph
🎯 What it does: The study addresses the fairness issue of allocating indivisible items in d-claw-free graphs, proving that each agent can be allocated at least 1/d of their maximum minimax share (MMS) on such graphs, and provides approximation guarantees related to proportional fairness in more general cases.
APR: Online Distant Point Cloud Registration through Aggregated Point Cloud Reconstruction
Quan Liu (Shanghai Jiao Tong University), Minyi Guo (Shanghai Jiao Tong University)
Pose EstimationAutonomous DrivingConvolutional Neural NetworkAuto EncoderPoint Cloud
🎯 What it does: Proposed an APR framework based on autoencoders for online long-range point cloud registration, leveraging aggregated point cloud reconstruction to enhance feature representation;
Artificial Agents Inspired by Human Motivation Psychology for Teamwork in Hazardous Environments
Anupama Arukgoda (University of New South Wales), Kasun Gunawardana
Reinforcement LearningAgentic AI
🎯 What it does: Design a multi-agent team framework based on human motivation psychology (needs theory), endowing agents with three implicit motivations: power, achievement, and affinity. Employ probabilistic goal selection and dynamically switchable strategies, evaluating their collaborative performance in fire-fighting tasks.
Asynchronous Communication Aware Multi-Agent Task Allocation
Ben Rachmut (Ben Gurion University of the Negev), Roie Zivan (Ben Gurion University of the Negev)
OptimizationTabular
🎯 What it does: Proposes an asynchronous communication-aware multi-agent task allocation algorithm called FMC ATA, aimed at solving task allocation problems with spatial and temporal constraints in physical environments, particularly suitable for scenarios with unreliable communication and dynamically changing environments.
Augmenting Automated Spectrum Based Fault Localization for Multiple Faults
Prantik Chatterjee (Indian Institute of Technology Kanpur), Subhajit Roy (Indian Institute of Technology Kanpur)
TabularBenchmark
🎯 What it does: Proposed ARTEMIS, an SBFL enhancement method based on multi-universe analysis, which improves the effectiveness of multi-defect localization by simulating the repair of dominant defects.
Auto-bidding with Budget and ROI Constrained Buyers
Xiaodong Liu (Renmin University of China), Weiran Shen (Renmin University of China)
OptimizationTabularSequentialFinance Related
🎯 What it does: This paper designs an automatic bidding system applicable to second-price auctions where buyers have private budgets and ROI constraints. It constructs a non-convex optimization model and provides a solution algorithm. It also proves that reporting true budgets and ROI is optimal for buyers under this system, discusses Bayesian Nash equilibrium and the lower bound of social welfare, and validates the effectiveness of the design through simulations.
Automatic Truss Design with Reinforcement Learning
Weihua Du (Tsinghua University), Yi Wu (Tsinghua University)
OptimizationGraph Neural NetworkTransformerReinforcement LearningGraphBenchmark
🎯 What it does: Propose AutoTruss, a two-stage method that first uses UCT to search for diverse feasible triangular truss layouts, then employs Soft Actor-Critic for fine-grained reinforcement learning refinement of node positions and cross-sectional dimensions.
Automatic Verification for Soundness of Bounded QNP Abstractions for Generalized Planning
Zhenhe Cui (Sun Yat-sen University), Yongmei Liu (Sun Yat-sen University)
🎯 What it does: Provide a proof-theoretic characterization of the soundness of general planning (GP) abstraction, propose automatically verifiable sufficient conditions, and subsequently implement a Bounded QNP abstraction soundness verification system based on an SMT solver.
Autonomous Exploration for Navigating in MDPs Using Blackbox RL Algorithms
Pratik Gajane (Eindhoven University of Technology), Ronald Ortner (Montanuniversitat Leoben)
Meta LearningReinforcement Learning
🎯 What it does: Propose a general meta-algorithm META-EXPLORE that converts any online RL algorithm with a sublinear regret upper bound into an autonomous motivation exploration algorithm, and provides upper bounds on sample complexity.
Backpropagation of Unrolled Solvers with Folded Optimization
James Kotary (University of Virginia), Ferdinando Fioretto (University of Virginia)
OptimizationComputational EfficiencyImageFinance Related
🎯 What it does: This paper proposes a folded optimization framework, which transforms the backpropagation of optimization layers embedded in deep networks from traditional unrolling to solving a linear system at known optimal points, significantly reducing computational overhead and improving accuracy.
BARA: Efficient Incentive Mechanism with Online Reward Budget Allocation in Cross-Silo Federated Learning
Yunchao Yang (Sun Yat-sen University), Quan Z. Sheng (Macquarie University)
OptimizationFederated LearningConvolutional Neural NetworkImage
🎯 What it does: Designed and implemented an online reward budget allocation algorithm called BARA to dynamically schedule reward budgets per round in cross-silo federated learning, aiming to improve the final model accuracy.
Basket Representation Learning by Hypergraph Convolution on Repeated Items for Next-basket Recommendation
Yalin Yu (Northeastern University), Xingwei Wang (Northeastern University)
Recommendation SystemRecurrent Neural NetworkGraph Neural NetworkContrastive LearningSequential
🎯 What it does: This paper proposes a basket representation learning framework (BRL) based on hypergraph convolution and contrastive learning to improve the accuracy of next-basket recommendation.
Bayesian Optimization with Switching Cost: Regret Analysis and Lookahead Variants
Peng Liu (Singapore Management University), Wei Qiyu (Shanghai University)
OptimizationHyperparameter SearchTabularBenchmark
🎯 What it does: Proposed a distance-adjusted Bayesian optimization method called distUCB, which simultaneously considers the objective function value and path switching costs during the search process to achieve more energy-efficient global search;
Beyond Homophily: Robust Graph Anomaly Detection via Neural Sparsification
Zheng Gong (University of Science and Technology of China), Jingyu Peng (University of Science and Technology of China)
Anomaly DetectionGraph Neural NetworkGraph
🎯 What it does: Propose a graph anomaly detection framework called SparseGAD based on neural sparsification, which removes structural noise and combines homogeneity and heterogeneity information for robust anomaly detection.
Beyond Pure Text: Summarizing Financial Reports Based on Both Textual and Tabular Data
Ziao Wang (Harbin Institute of Technology), Hongwei Du (Harbin Institute of Technology)
TransformerLarge Language ModelMultimodalityTabularFinance RelatedRetrieval-Augmented Generation
🎯 What it does: Proposed a model that simultaneously utilizes tabular data and textual data to generate financial report summaries, and manually constructed a specialized 'table + text → summary' dataset.
Beyond Strict Competition: Approximate Convergence of Multi-agent Q-Learning Dynamics
Aamal Hussain (Imperial College London), Georgios Piliouras (Singapore University of Technology and Design)
Reinforcement LearningGraph
🎯 What it does: Investigate the approximate convergence behavior of smooth Q-learning (with Boltzmann exploration) in non-strict zero-sum but 'approximate zero-sum' network games, proving its convergence to the neighborhood of the unique QRE.
Bi-level Dynamic Learning for Jointly Multi-modality Image Fusion and Beyond
Zhu Liu (Dalian University Of Technology), Risheng Liu (Dalian University Of Technology)
Object DetectionSegmentationOptimizationConvolutional Neural NetworkGenerative Adversarial NetworkImageMultimodality
🎯 What it does: This paper proposes a dual-layer dynamic learning framework for jointly optimizing multi-modal (infrared + visible) image fusion and downstream semantic perception tasks (object detection and semantic segmentation).
Bidirectional Dilation Transformer for Multispectral and Hyperspectral Image Fusion
Shangqi Deng (University of Electronic Science and Technology of China), Rui Wen (University of Electronic Science and Technology of China)
TransformerImage
🎯 What it does: A bidirectional dilated Transformer (BDT) architecture is proposed for the fusion of multispectral and hyperspectral images.
Bipolar Abstract Dialectical Frameworks Are Covered by Kleene’s Three-valued Logic
Ringo Baumann (Leipzig University), Maximilian Heinrich (Bauhaus University Weimar)
Computational Efficiency
🎯 What it does: This paper demonstrates that Kleene's three-valued logic can be directly applied in Bipolar Abstract Dialogue Frameworks (BADFs) to compute the Γ-operator, thereby avoiding the enumeration of all two-valued completions.
Black-Box Data Poisoning Attacks on Crowdsourcing
Pengpeng Chen (China's Aviation System Engineering Research Institute), Peng Lin (China's Aviation System Engineering Research Institute)
Adversarial AttackText
🎯 What it does: This paper proposes a SubPac framework for black-box attacks, which can achieve data destruction by optimizing instance selection and label poisoning under unknown label aggregation models.
Black-box Prompt Tuning for Vision-Language Model as a Service
Lang Yu (East China Normal University), Liang He (East China Normal University)
ClassificationOptimizationPrompt EngineeringVision Language ModelImageTextMultimodality
🎯 What it does: Propose a black-box prompt tuning framework BPT-VLM for vision-language models in the MaaS scenario, which can learn task-related visual and linguistic prompts without accessing gradients or model structures;
Boosting Decision-Based Black-Box Adversarial Attack with Gradient Priors
Han Liu (Dalian University of Technology), Xianchao Zhang (Dalian University of Technology)
Adversarial AttackConvolutional Neural NetworkImage
🎯 What it does: This study proposes a decision-based black-box attack framework (DBA-GP) based on gradient prior, improving gradient estimation by combining joint bilateral filtering and time-related gradient updates;
Boosting Few-Shot Open-Set Recognition with Multi-Relation Margin Loss
Yongjuan Che (Southeast University), Hui Xue (Southeast University)
ClassificationRecognitionConvolutional Neural NetworkTransformerImageBenchmark
🎯 What it does: This paper proposes a Multi-Relationship Margin Loss (MRM) to significantly enhance the detection capability of unknown samples in few-shot open set recognition (FSOSR) tasks while maintaining the classification accuracy of known classes.
BPNet: Bézier Primitive Segmentation on 3D Point Clouds
Rao Fu (Inria), Pierre Alliez (Inria)
SegmentationComputational EfficiencyGraph Neural NetworkPoint Cloud
🎯 What it does: Propose BPNet, an end-to-end deep learning framework for segmenting 3D point clouds into B-spline (Bézier) primitives.
BRExIt: On Opponent Modelling in Expert Iteration
Daniel Hernandez (Sony AI), Michael Kaisers (Centrum Wiskunde & Informatica)
Reinforcement Learning from Human FeedbackReinforcement LearningSequential
🎯 What it does: Proposed a reinforcement learning framework named BRExIt, which achieves faster learning of optimal counter-strategies by incorporating opponent modeling into Expert Iteration (ExIt).
Building Concise Logical Patterns by Constraining Tsetlin Machine Clause Size
K. Darshana Abeyrathna (University of Agder), Xuan Zhang (Norwegian Research Centre)
Explainability and InterpretabilityComputational EfficiencyImageText
🎯 What it does: This paper proposes Clause Size Constrained Tsetlin Machine (CSC-TM), which introduces a soft constraint on clause length during the learning process of Tsetlin Machine to limit the number of literals in each clause, thereby obtaining more concise and interpretable logical rules.
c-TPE: Tree-structured Parzen Estimator with Inequality Constraints for Expensive Hyperparameter Optimization
Shuhei Watanabe (University of Freiburg), Frank Hutter (University of Freiburg)
OptimizationHyperparameter SearchNeural Architecture SearchTabularBenchmark
🎯 What it does: Proposed a constrained version of TPE, called c-TPE, based on the tree-structured Parzen estimator (TPE), for addressing inequality constraints in hyperparameter optimization.
CADParser: A Learning Approach of Sequence Modeling for B-Rep CAD
Shengdi Zhou (Beihang University), Bin Zhou (Beihang University)
Graph Neural NetworkTransformerGraphSequential
🎯 What it does: Utilize deep learning to infer the construction sequence of B-Rep CAD models, and propose the CADParser network to achieve automatic parsing of CAD construction steps.
Calibrating a Deep Neural Network with Its Predecessors
Linwei Tao (University of Sydney), Chang Xu (University of Sydney)
ClassificationExplainability and InterpretabilityImage
🎯 What it does: Proposed a predecessor combination search (PCS) method that achieves better model calibration by selecting the optimal predecessor weight combinations for each network block.
Can I Really Do That? Verification of Meta-Operators via Stackelberg Planning
Florian Pham (Saarland University), Alvaro Torralba (Aalborg University)
OptimizationMeta LearningBenchmark
🎯 What it does: Propose and verify a new meta-operator that can achieve the desired effect through different action sequences under any state satisfying the preconditions, and determine its validity by compiling it into a Stackelberg plan in one go.
Can You Improve My Code? Optimizing Programs with Local Search
Fatemeh Abdollahi (University of Alberta), Levi H. S. Lelis (University of Alberta)
OptimizationText
🎯 What it does: This paper proposes a program optimization method called POLIS, which uses local search and existing enumerative synthesizers to incrementally improve existing programs line by line, aiming to enhance quantifiable objective functions (e.g., game scores)
Capturing the Long-Distance Dependency in the Control Flow Graph via Structural-Guided Attention for Bug Localization
Yi-Fan Ma (Nanjing University), Ming Li (Nanjing University)
AI Code AssistantGraph Neural NetworkTransformerLarge Language ModelTextGraph
🎯 What it does: This paper proposes a model called sgAttention that utilizes structural guided attention to capture long-range dependencies in control flow graphs (CFGs), thereby enhancing defect localization effectiveness.
Cardinality-Minimal Explanations for Monotonic Neural Networks
Ouns El Harzli (University of Oxford), Ian Horrocks (University of Oxford)
Explainability and InterpretabilityComputational EfficiencyTabularFinance Related
🎯 What it does: Studied the cardinality minimization and inductive explanation problems for monotonic neural networks (with differentiable continuous activation functions), and proposed a polynomial-time greedy algorithm to compute these explanations;
Case-Based Reasoning with Language Models for Classification of Logical Fallacies
Zhivar Sourati (University of Southern California), Alain Mermoud (armasuisse Science and Technology)
ClassificationTransformerLarge Language ModelPrompt EngineeringTextRetrieval-Augmented Generation
🎯 What it does: The study proposes a method that combines case-based reasoning (CBR) with language models to classify logical fallacies in natural language arguments.
Causal Deep Reinforcement Learning Using Observational Data
Wenxuan Zhu (Dalian University of Technology), Qiang Zhang (Dalian University of Technology)
Reinforcement LearningTabularBenchmark
🎯 What it does: This paper proposes two deconfounding methods based on causal inference (reweighting and resampling) to remove the influence of unobserved confounding variables in observational data for offline deep reinforcement learning;
Causal-Based Supervision of Attention in Graph Neural Network: A Better and Simpler Choice towards Powerful Attention
Hongjun Wang (Southern University of Science and Technology), Xuan Song (Southern University of Science and Technology)
Graph Neural NetworkGraph
🎯 What it does: Proposes a causal inference-based attention supervision framework named CSA, which directly maximizes the causal effect of attention on predictions, thereby improving the attention mechanism in graph neural networks.
Character As Pixels: A Controllable Prompt Adversarial Attacking Framework for Black-Box Text Guided Image Generation Models
Ziyi Kou (University of Notre Dame), Xiangliang Zhang (University of Notre Dame)
Adversarial AttackTransformerVision Language ModelImageText
🎯 What it does: Studied controllable character-level adversarial attacks on black-box text-to-image generation models and proposed the CharGrad framework.
Choose your Data Wisely: A Framework for Semantic Counterfactuals
Edmund Dervakos (National Technical University of Athens), Giorgos Stamou (National Technical University of Athens)
Explainability and InterpretabilityData-Centric LearningImageAudio
🎯 What it does: Proposed a semantic inverse causal explanation framework based on knowledge graphs, along with a corresponding algorithm for computing the minimal edit set.
Choosing Well Your Opponents: How to Guide the Synthesis of Programmatic Strategies
Rubens O. Moraes (Universidade Federal de Viçosa), Levi H. S. Lelis (University of Alberta)
Computational EfficiencyReinforcement LearningBenchmark
🎯 What it does: Propose the Local Learner (2L) algorithm, which synthesizes readable procedural strategies in two-player zero-sum games using the PSRO framework and local search (Hill-Climbing).