AAAI 2026 Papers — Page 29
AAAI Conference on Artificial Intelligence · 4149 papers
Planning in Branch-and-Bound: Model-Based Reinforcement Learning for Exact Combinatorial Optimization
Paul Strang (Edf R&D), Emmanuel Rachelson
OptimizationGraph Neural NetworkReinforcement LearningTabularBenchmark
🎯 What it does: Propose PlanB&B, a model-based reinforcement learning agent that improves MILP variable selection strategies by learning a Branch-and-Bound (B&B) dynamic model and performing planning.
Planning with Uncertain Action Models
Francesco Percassi (University of Huddersfield), Enrico Scala (University of Brescia)
OptimizationBenchmark
🎯 What it does: Propose the PUMA (Planning with Uncertain Models of Actions) framework, where action models are revealed after the first execution and become deterministic in subsequent executions, and present two polynomial compilation methods: one compiled into FOND (COMP2FOND), and the other compiled into classical planning (COMP2FOD).
PLaST: Towards Paralinguistic-aware Speech Translation
Yi Li (Xiamen University), Yidong Chen (Huawei Translation Services Center)
TransformerLarge Language ModelTextMultimodalityAudio
🎯 What it does: Proposed a dual-branch end-to-end speech translation framework called PLaST, specifically designed to simultaneously capture linguistic content in speech and non-linguistic cues such as emotion, stress, and integrate them into large language models (LLMs) for translation;
Plot’n Polish: Zero-Shot Story Visualization and Disentangled Editing with Text-to-Image Diffusion Models
Kiymet Akdemir (Virginia Tech), Pinar Yanardag (Virginia Tech)
GenerationConvolutional Neural NetworkTransformerLarge Language ModelPrompt EngineeringVision Language ModelDiffusion modelImageTextMultimodality
🎯 What it does: Propose a zero-training story visualization and multi-frame editing framework Plot'n Polish, which supports fine-grained and global editing of an entire set of story images without regenerating them.
Plug-and-Play Clarifier: A Zero-Shot Multimodal Framework for Egocentric Intent Disambiguation
Sicheng Yang, Zhensong Zhang (Imperial College London)
Object DetectionSegmentationDepth EstimationTransformerLarge Language ModelVision Language ModelImageTextMultimodalityChain-of-Thought
🎯 What it does: Propose Plug-and-Play Clarifier, a zero-shot, modular framework that progressively eliminates modality ambiguity in front-view interactions through three clarification modules: text, visual, and cross-modal, thereby improving intent understanding in egocentric AI agents.
Plug-and-Play Optimization for 3D Gaussian Splatting Compression: Distribution Regularization, Probabilistic Pruning and Detail Compensation
Tian Bai (University of Science and Technology of China), Ziyang Dai (University of Science and Technology of China)
CompressionGaussian Splatting
🎯 What it does: This paper proposes a three-module plug-and-play optimization framework (distribution regularization, probabilistic pruning, and high-frequency compensation) to significantly reduce the storage and computational costs of 3D Gaussian Splatting (3DGS) models, while seamlessly integrating into mainstream structured compression methods.
Plug-and-Play Parameter-Efficient Tuning of Embeddings for Federated Recommendation
Haochen Yuan (Harbin Institute of Technology), Zhongjie Wang (Harbin Institute of Technology)
Recommendation SystemFederated LearningAuto EncoderTabular
🎯 What it does: Propose a pluggable federated recommendation framework that adopts parameter-efficient fine-tuning (PEFT) to update only compressed item embeddings on the client side, thereby significantly reducing communication overhead;
PlugTrack: Multi-Perceptive Motion Analysis for Adaptive Fusion in Multi-Object Tracking
Seungjae Kim (Kyung Hee University), MyeongAh Cho (Kyung Hee University)
Object TrackingRecurrent Neural NetworkVideo
🎯 What it does: Propose the PlugTrack framework, which achieves adaptive fusion of Kalman filters and data-driven motion predictors through multi-sensory motion analysis;
PLUM-Net: Prototype-Induced Label Structuring for Disentangled Multimodal Representation Network
Kehan Wang (Hunan University), Zixing Zhang (Hunan University)
Representation LearningTransformerContrastive LearningMultimodality
🎯 What it does: Proposes PLUM-Net, a multimodal representation network that explicitly separates and fuses shared and private features through prototype-induced label generation and multi-layer semantic alignment.
PMGS: Reconstruction of Projectile Motion Across Large Spatiotemporal Spans via 3D Gaussian Splatting
Yijun Xu (Wuhan University), Chu He (Chongqing University)
Pose EstimationGaussian SplattingOptical FlowVideoPhysics Related
🎯 What it does: This work proposes the PMGS framework, which reconstructs the complete spatiotemporal trajectories of projected motion from monocular videos using 3D Gaussian splatting.
PMPGuard: Catching Pseudo-Matched Pairs in Remote Sensing Image–Text Retrieval
Pengxiang Ouyang, Cong Bai (Zhejiang University Of Technology)
RetrievalTransformerContrastive LearningImageText
🎯 What it does: Proposed the PMPGuard framework, which utilizes cross-modal gated attention and positive-negative awareness attention to handle pseudo-matched pairs in remote sensing image-text retrieval.
PocketLLM: Ultimate Compression of Large Language Models via Meta Networks
Ye Tian (Huawei Noah's Ark Lab), Kai Han (University of Sydney)
CompressionComputational EfficiencyTransformerLarge Language ModelText
🎯 What it does: This study proposes the PocketLLM method, which projects LLM weights into a latent space via a meta-encoder, quantizes them using a discrete codebook, and reconstructs the weights with a meta-decoder, achieving ultra-high compression ratios (10×–20×).
PoeTone: A Framework for Constrained Generation of Structured Chinese Songci with LLMs
Zhan Qu (TU Dresden), Michael Färber
GenerationTransformerLarge Language ModelSupervised Fine-TuningPrompt EngineeringText
🎯 What it does: This paper studies the performance of large language models in generating Song dynasty lyrics under structured constraints, and constructs the PoeTone evaluation framework and the Generate-Critic fine-tuning method.
Point Cloud Quality Assessment via Multi-View Structure-Aware Feature Fusion
Jian Xiong (Nanjing University of Posts and Telecommunications), Hao Gao (Nanjing University of Posts and Telecommunications)
Convolutional Neural NetworkTransformerPoint Cloud
🎯 What it does: This paper proposes a multi-view structure-aware feature fusion network (SAF-Net) for no-reference point cloud quality assessment. The network generates texture maps, object mask maps, and local binary pattern (LBP) maps through projection, extracts geometric and texture information separately, and predicts subjective quality scores via a hybrid CNN-ViT encoder and two-stage Transformer fusion.
Point Cloud Quantization Through Multimodal Prompting for 3D Understanding
Hongxuan Li (Tianjin University), Pengfei Zhu (Zhejiang Normal University)
ClassificationRecognitionSegmentationTransformerPrompt EngineeringVision Language ModelTextMultimodalityPoint Cloud
🎯 What it does: Proposes a text-prompt based point cloud quantization framework, which discretizes continuous point cloud features into text-driven prototypes and fuses visual details with semantic information through cross-attention.
Point Cloud Segmentation of Integrated Circuits Package Substrates Surface Defects Using Causal Inference: Dataset Construction and Methodology
Bingyang Guo (Northeastern University), Ruiyun Yu (Northeastern University)
SegmentationAnomaly DetectionTransformerPoint CloudPhysics Related
🎯 What it does: Developed a high-resolution point cloud dataset of surface defects on ceramic packaging substrates named CPS3D-Seg, and proposed a 3D segmentation network based on causal inference called CINet.
Point Cloud Semantic Scene Completion with Prototype-Guided Transformer
Chenghao Fang, Feilong Cao (Shanxi University)
SegmentationConvolutional Neural NetworkTransformerPoint Cloud
🎯 What it does: Propose ProtoFormer, a Transformer-based point cloud semantic scene completion method that simultaneously accomplishes point cloud reconstruction and semantic annotation by leveraging learnable semantic prototypes and top-K attention mechanisms.
Point-SRA: Self-Representation Alignment for 3D Representation Learning
Lintong Wei, Kaibing Zhang (Xi'an Jiaotong University)
Object DetectionSegmentationRepresentation LearningTransformerFlow-based ModelAuto EncoderContrastive LearningPoint CloudBiomedical DataBenchmark
🎯 What it does: This paper proposes Point-SRA, a 3D self-supervised representation learning framework that combines mask ratio complementarity with MeanFlow probabilistic modeling.
PointChain: Learning Generalizable Point Cloud Representations via Structural Chain Modeling
Luyao Wang (University of Science and Technology of China), Tianzhu Zhang (University of Science and Technology of China)
Representation LearningTransformerContrastive LearningPoint Cloud
🎯 What it does: Proposed PointChain, a self-regressive point cloud pre-training framework based on structural chain encoding.
PointDGRWKV: Generalizing RWKV-like Architecture to Unseen Domains for Point Cloud Classification
Hao Yang (Shanghai Jiao Tong University), Shuicheng YAN (National University of Singapore)
ClassificationDomain AdaptationTransformerPoint Cloud
🎯 What it does: This paper proposes a domain generalization model called PointDGRWKV for point cloud classification, addressing the issues of local geometry modeling and cross-domain attention drift in RWKV on point cloud tasks.
PointMC: Multi-view Consistent Encoding and Center-Global Feature Fusion for Point Clouds Understanding
Xinxing Yu (Macau University of Science and Technology), Yanyan Liang (Macau University of Science and Technology)
ClassificationSegmentationRepresentation LearningTransformerPoint CloudBenchmark
🎯 What it does: A new framework called PointMC is proposed for point cloud tasks, enhancing point cloud understanding performance through multi-view consistent learnable position encoding (MCLPE) and center-global feature fusion (CGFF).
Points Meet Pixels: Bridging 2D Vision-Language Model and 3D Perception Gaps for Point Cloud Quality Assessment
Mingxuan Li (Beijing Institute of Technology), Runze Hu (Beijing Institute of Technology)
Vision Language ModelPoint Cloud
🎯 What it does: Proposed PMP-PCQA, a point cloud quality assessment framework based on vision-language models, which bridges 2D VLM and 3D point clouds through point-pixel fine-grained correspondence, and constructs three modules (SAE, FCA, TAM) to achieve spatial enhancement, cross-modal consistency, and quality-sensitive feature mining.
PointSLAM++: Robust Dense Neural Gaussian Point Cloud-based SLAM
Xu Wang (Hunan University), Ruihui Li (Hunan University)
Gaussian SplattingSimultaneous Localization and MappingPoint Cloud
🎯 What it does: Proposed a real-time RGB-D SLAM system called PointSLAM++, achieving precise 3D reconstruction and photorealistic rendering through hierarchical neural Gaussian representations and hierarchical pose optimization.
Poisoned Distillation: Injecting Backdoors into Distilled Datasets Without Raw Data Access
Ziyuan Yang (Sichuan University), Joey Tianyi Zhou (Agency for Science, Technology and Research)
Knowledge DistillationAdversarial AttackConvolutional Neural NetworkImage
🎯 What it does: Proposes an attack method to inject backdoors into distilled datasets without requiring access to the original data.
Poisoning with a Pill: Circumventing Detection in Federated Learning
Hanxi Guo (Purdue University), Xiangyu Zhang (Queen's University Belfast)
Federated LearningAdversarial AttackImage
🎯 What it does: Designed an attack enhancement method based on a 'pill' subnetwork, which can focus existing poisoning attacks on critical subnetworks of the model in federated learning, thereby enhancing the attack effectiveness while maintaining stealthiness.
Polarization Uncertainty-Guided Diffusion Model for Color Polarization Image Demosaicking
Chenggong Li (Central South University), Degui Yang (Central South University)
RestorationConvolutional Neural NetworkTransformerDiffusion modelImage
🎯 What it does: Propose a color polarization image demosaicking method based on polarization uncertainty guided diffusion models.
Policy Newton Methods for Distortion Riskmetrics
Soumen Pachal (Indian Institute of Technology Madras), Prashanth L. A. (Indian Institute of Technology Madras)
OptimizationReinforcement LearningBenchmark
🎯 What it does: Proposed a policy Newton algorithm for Distortion Risk Measure (DRM), utilizing policy Hessian matrix estimation to achieve risk-sensitive reinforcement learning.
Policy Search, Retrieval, and Composition via Task Similarity in Collaborative Agentic Systems
Saptarshi Nath (Loughborough University), Andrea Soltoggio (Loughborough University)
Reinforcement LearningAgentic AIBenchmarkRetrieval-Augmented Generation
🎯 What it does: Propose the MOSAIC algorithm, which helps agents select, integrate, and fine-tune strategies learned by others in distributed environments to accelerate their own learning.
Policy Zooming: Adaptive Discretization-based Infinite-Horizon Average-Reward Reinforcement Learning
Avik Kar (Indian Institute of Science), Rahul Singh (Indian Institute of Science)
OptimizationReinforcement Learning
🎯 What it does: Studied average reward reinforcement learning in continuous space Lipschitz MDPs, and proposed adaptive exploration algorithms PZRL-MF and PZRL-MB based on policy zooming, providing finite-time regret upper bounds;
Polysemic Semantic Instance Network for Cross-Modal Hashing
Shuo Han, Lei Huang (Qingdao University of Technology)
RetrievalTransformerMultimodality
🎯 What it does: Propose a multi-instance multi-modal hashing framework called DPSIH, which generates various semantic embeddings through multi-head self-attention and residual learning to achieve cross-modal retrieval.
PortraitSR: Artist-Inspired Prior Learning for Progressive Face Super-Resolution
Miaoqing Wang (Chongqing University of Post and Telecommunication), Long Sun (Nanjing University of Science and Technology)
Super ResolutionTransformerImage
🎯 What it does: Proposed a facial super-resolution framework called PortraitSR inspired by human painting, which includes structure-priority, texture-priority, and holistic fusion modules to recover high-quality facial images from low-resolution inputs.
Position Fair Mechanisms Allocating Indivisible Goods
Ryoga Mahara, Tomohiko Yokoyama (University Of Tokyo)
🎯 What it does: This paper proposes a new concept of mechanism fairness for indivisible items—Positional Equilibrium to One Item (PEF1)—and designs polynomial-time, scale-invariant mechanisms that guarantee EF1 allocations; for two-person cases, it also proves that the Maximum Nash Welfare (MNW) and Adjusted Winner mechanisms satisfy PEF1 and can achieve EF1 and Pareto optimal allocations; meanwhile, it analyzes the PEF1 performance of classic round-robin and envy-cycle mechanisms.
Positional Bias in Multimodal Embedding Models: Do They Favor the Beginning, the Middle, or the End?
Kebin Wu (Technology Innovation Institute), Fatima Albreiki (Technology Innovation Institute)
ClassificationRetrievalTransformerVision Language ModelContrastive LearningImageTextMultimodality
🎯 What it does: Systematic evaluation and quantification of positional information bias in multimodal representation models (e.g., CLIP and its variants) for image-text retrieval tasks, distinguishing positional bias from contextual importance.
Positional Cognitive Specialization: Where Do LLMs Learn to Comprehend and Speak Your Language?
Luis Frentzen Salim (Academia Sinica), Hsing-Kuo Kenneth Pao (Academia Sinica)
Computational EfficiencyRepresentation LearningTransformerLarge Language ModelSupervised Fine-TuningText
🎯 What it does: Investigated the specialization of perception (understanding) and production (generation) functions in the front and back layers of large language models (LLMs) when learning low-resource languages, and proposed a layer selection strategy called CogSym, which trains only the outer 25% of layers, significantly reducing adaptation costs.
Positive Definite Sparse Covariance Estimation via Dual Space Optimization
Fengpei Li (ShanghaiTech University), Ziping Zhao (ShanghaiTech University)
OptimizationComputational EfficiencyTabularFinance Related
🎯 What it does: This paper revisits the problem of positive definite sparse covariance estimation (PDSCE) and proposes a new dual proximal gradient method (DPGM) to address this issue.
PosPrune: Visual Token Pruning with Positional Bias Correction for Efficient Large Vision-Language Models
Ziyang Wang (University of Science and Technology of China), Zilei Wang (University of Science and Technology of China)
Computational EfficiencyVision Language ModelMultimodalityBenchmark
🎯 What it does: Designed and implemented a training-free two-stage visual token pruning framework called PosPrune, which first adaptively prunes tokens based on image semantic density in the visual encoder, and then corrects positional bias in the LLM decoder while employing MMR for text-guided diverse pruning, significantly reducing the number of visual tokens and accelerating inference.
Post Training Quantization for Efficient Dataset Condensation
Linh-Tam Tran (Kyung Hee University), Sung-Ho Bae (Kyung Hee University)
CompressionComputational EfficiencyConvolutional Neural NetworkImage
🎯 What it does: This paper proposes a post-training quantization scheme to compress synthetic samples to extremely low bit-widths (e.g., 2-bit) during dataset condensation while maintaining training effectiveness.
Post-Hoc Refinement for Multitask Symbolic Regression via Consensus-Accelerated Shapley Analysis
Xinyue Li (University of Electronic Science and Technology of China), Yu Zhang (University of Electronic Science and Technology of China)
OptimizationExplainability and InterpretabilityComputational EfficiencyRepresentation LearningTabularBenchmark
🎯 What it does: This paper proposes an MTGP-BS framework that synthesizes higher quality expressions by refining and reconstructing sub-expressions of the entire population in the later stages of multi-task symbolic regression.
Posterior Label Smoothing for Node Classification
Jaeseung Heo (POSTECH), Dongwoo Kim (POSTECH)
ClassificationGraph Neural NetworkGraph
🎯 What it does: Propose a node classification method based on posterior label smoothing, which adaptively generates soft labels by leveraging neighbor labels and global statistics.
PosterVerse: A Full-Workflow Framework for Commercial-Grade Poster Generation with HTML-Based Scalable Typography
Junle Liu (South China University of Technology), Lianwen Jin (South China University of Technology)
GenerationTransformerLarge Language ModelSupervised Fine-TuningPrompt EngineeringVision Language ModelDiffusion modelImageTextMultimodality
🎯 What it does: Built a full-flow commercial-grade poster generation framework called PosterVerse, which supports automatically generating blueprints, background images, and implementing editable, scalable text layouts using HTML; simultaneously launched the PosterDNA Chinese poster generation dataset, which includes three subsets: blueprints, backgrounds, and HTML.
Potent but Stealthy: Rethink Profile Pollution Against Sequential Recommendation via Bi-Level Constrained Reinforcement Paradigm
Jiajie Su (Zhejiang University), Chaochao Chen (Zhejiang University)
Recommendation SystemOptimizationAdversarial AttackRecurrent Neural NetworkTransformerReinforcement LearningSequential
🎯 What it does: This paper proposes a stealthy profile pollution attack (CREAT) based on constrained reinforcement learning, achieving precise misguidance of sequential recommenders through dual-layer optimization and pattern balance reward strategies.
Potential Outcome Rankings for Counterfactual Decision Making
Yuta Kawakami (Mohamed bin Zayed University of Artificial Intelligence), Jin Tian (Mohamed bin Zayed University of Artificial Intelligence)
TabularBiomedical Data
🎯 What it does: Proposed and studied two new causal decision metrics—Potential Outcome Rank Probability (PoR) and Best Outcome Probability (PoB), and provided their identification theorems, estimation methods, and bounds.
PPFL: A Parameter Behavior-Driven Plug-in Personalization Engine for Federated Learning
Qianyue Cao (University of Science and Technology of China), Xuehai Zhou (University of Science and Technology of China)
Federated LearningImageTextMultimodality
🎯 What it does: Proposed a dynamic soft fusion plugin called PPFL based on parameter behavior perception for personalized federated learning, aiming to address the inherent drift problem caused by traditional model decoupling.
PPGPT: Transferring Next-Token Modeling from Language to PPG Signals
Zexing Zhang (Changchun University of Technology), Qingxin Zhao (Changchun University of Technology)
ClassificationTransformerLarge Language ModelMixture of ExpertsContrastive LearningBiomedical DataBenchmark
🎯 What it does: Built a PPG-based foundation model called PPGPT, pre-trained using Next-Feature Token Prediction to achieve multi-task PPG signal analysis.
PQDA:Policy-Aligned Q-Consistency Meets Decoupled Augmentation for Generalizable Visual RL
Yun Zhou (Anhui University), Chunyu Tan (Anhui University)
Reinforcement LearningContrastive LearningImageVideoBenchmark
🎯 What it does: This paper proposes the PQDA framework to address the generalization challenges of visual reinforcement learning in visually interfering environments, integrating foreground-background separation augmentation with strategy-aligned Q consistency, while removing auxiliary tasks.
Practical Global and Local Bounds in Gaussian Process Regression via Chaining
Junyi Liu (National University of Singapore), Stanley Kok (National University of Singapore)
Explainability and InterpretabilityTabular
🎯 What it does: Proposed a framework based on the chaining method for estimating upper and lower bounds on the global expected maximum and local uncertainty intervals in Gaussian Process Regression (GPR).
Practical, Utilitarian Algorithm Configuration
Devon R. Graham (University of British Columbia), Kevin Leyton-Brown (University of British Columbia)
OptimizationHyperparameter SearchBenchmark
🎯 What it does: This paper improves the COUP algorithm for configuration methods, significantly enhancing practical performance while maintaining theoretical optimality guarantees, and demonstrates its effectiveness in SAT solution configuration.
PrAda-GAN: A Private Adaptive Generative Adversarial Network with Bayes Network Structure
Ke Jia (Renmin University of China), Feifei Wang (Renmin University of China)
Data SynthesisSafty and PrivacyGenerative Adversarial NetworkTabular
🎯 What it does: Propose a differential privacy table data generation model PrAda-GAN that combines GAN with Bayesian network structure, using a sequential generator and sparse regularization to approximate the real data distribution;
PRAGWORLD: A Benchmark Evaluating LLMs’ Local World Model Under Minimal Linguistic Alterations and Conversational Dynamics
Sachin Vashistha (Indian Institute of Technology Kharagpur), Somak Aditya (Indian Institute of Technology Kharagpur)
Explainability and InterpretabilityTransformerLarge Language ModelTextBenchmark
🎯 What it does: Evaluate the implicit world model plasticity of LLMs in dialogues, construct the PRAGWORLD benchmark, and test model robustness through seven minimal language variants.
Pre-DPO: Improving Data Utilization in Direct Preference Optimization Using a Guiding Reference Model
Junshu Pan (Zhejiang University), Yue Zhang (Westlake University)
OptimizationHyperparameter SearchReinforcement Learning from Human FeedbackTransformerLarge Language ModelSupervised Fine-TuningReinforcement LearningText
🎯 What it does: Propose Pre-DPO, a paradigm that introduces a guiding reference model in DPO/SimPO training. It first performs a standard preference optimization on the policy, then uses the optimized model as a reference to retrain the original policy, thereby better weighting the training samples.
Pre-Trained Video Generative Models as World Simulators
Haoran He (Hong Kong University Of Science And Technology), Ling Pan (Tencent Ai Lab)
GenerationTransformerSupervised Fine-TuningReinforcement LearningDiffusion modelFlow-based ModelVideo
🎯 What it does: Proposed a framework called DWS, which can convert pre-trained large-scale video generation models into world simulators controllable by action trajectories, achieving action-driven dynamic video generation.
Predict and Resist: Long-Term Accident Anticipation Under Sensor Noise
Xingcheng Liu (University of Macau), Zhenning Li (University of Macau)
Anomaly DetectionAutonomous DrivingOptimizationConvolutional Neural NetworkRecurrent Neural NetworkReinforcement LearningDiffusion modelVideoTime Series
🎯 What it does: Propose a framework combining diffusion denoising and time-aware actor-critic for proactive traffic accident prediction under sensor noise.
Predicting Emergent Tool Use in LLMs Before It Emerges: A Proxy Perspective
Bo-Wen Zhang, Xu-Cheng Yin (Beijing Academy Of Artificial Intelligence)
Explainability and InterpretabilityAI Code AssistantTransformerLarge Language ModelSupervised Fine-TuningTextBenchmarkRetrieval-Augmented Generation
🎯 What it does: Construct a proxy framework to predict the tool-use capability of large language models in the final training stage by evaluating the performance of a set of non-emergent proxy tasks early in training.
Predicting the Future by Retrieving the Past
Dazhao Du (Hong Kong University of Science and Technology), Song Guo (Hong Kong University of Science and Technology)
Contrastive LearningTime SeriesRetrieval-Augmented Generation
🎯 What it does: Proposes the PFRP (Predicting the Future by Retrieving the Past) framework, which explicitly leverages global historical data in time series prediction. It constructs a global memory bank (GMB) and retrieves relevant information, dynamically fusing the retrieved global predictions with the results of any local prediction model to enhance forecasting accuracy.
Predicting Video Slot Attention Queries from Random Slot-Feature Pairs
Rongzhen Zhao (Aalto University), Joni Pajarinen (Aalto University)
RecognitionObject DetectionTransformerAuto EncoderContrastive LearningVideoMultimodality
🎯 What it does: This paper proposes the RandSF.Q method, which uses random slot-feature pairs to improve query prediction in video unsupervised object-centric learning.
Preference Elicitation for Step-Wise Explanations in Logic Puzzles
Marco Foschini (KU Leuven), Tias Guns (KU Leuven)
OptimizationExplainability and InterpretabilityReinforcement LearningText
🎯 What it does: The study applies interactive preference elicitation (Constructive Preference Elicitation) to the step-by-step explanation of logic puzzles, aiming to learn users' linear preferences for explanation quality and generate more understandable explanation steps.
Preference Is More than Comparisons: Rethinking Dueling Bandits with Augmented Human Feedback
Shengbo Wang (University of Electronic Science and Technology of China), Ke Li (University of Exeter)
Recommendation SystemOptimizationReinforcement Learning from Human FeedbackReinforcement LearningBenchmark
🎯 What it does: Propose a model-free dueling bandit framework that achieves interactive preference acquisition through enhanced human feedback.
PrefixGPT: Prefix Adder Optimization by a Generative Pre-trained Transformer
Ruogu Ding (Shanghai Jiao Tong University), Weikang Qian (Shanghai Jiao Tong University)
OptimizationTransformerReinforcement LearningSequential
🎯 What it does: This study proposes PrefixGPT, a model based on generative pre-trained Transformers, capable of directly generating optimal prefix adder topologies from scratch that satisfy design rules.
Preserving Topological and Geometric Embeddings for Point Cloud Recovery
Kaiyue Zhou (Tsinghua University), Shengjin Wang (Tsinghua University)
RestorationTransformerPoint Cloud
🎯 What it does: Developed an end-to-end point cloud recovery framework called TopGeoFormer, which can simultaneously preserve topological and geometric features during the sampling and recovery process.
PressTrack-HMR: Pressure-Based Top-Down Multi-Person Global Human Mesh Recovery
Jiayue Yuan (University of Science and Technology of China), Xiaohui Cai (University of Science and Technology of China)
Object DetectionObject TrackingPose EstimationConvolutional Neural NetworkTransformerMeshTime SeriesBenchmark
🎯 What it does: Built an end-to-end pipeline for multi-person human mesh recovery from pressure pad data (PressTrack-HMR) and released the multi-person interactive pressure dataset MIP.
PRGB Benchmark: A Robust Placeholder-Assisted Algorithm for Benchmarking Retrieval-Augmented Generation
Zhehao Tan (Ant Group), Jinjie Gu (Ant Group)
RetrievalLarge Language ModelTextBenchmarkRetrieval-Augmented GenerationChain-of-Thought
🎯 What it does: Propose the Placeholder-RAG-Benchmark (PRGB) to conduct multi-dimensional fine-grained evaluation of LLMs' ability to utilize retrieved documents within RAG systems.
PriAgent: A Collaborative Multi-Agent Framework for Auditing Android Privacy Compliance
Ziwei Zhang (Chinese Academy of Sciences), Qingyun Liu (Chinese Academy of Sciences)
Safty and PrivacyExplainability and InterpretabilityComputational EfficiencyLarge Language ModelAgentic AIPrompt EngineeringTextBenchmarkRetrieval-Augmented GenerationChain-of-Thought
🎯 What it does: Designed and implemented PriAgent, a multi-agent AI framework that automatically completes compliance audits for Android app code and natural language privacy policies, forming a complete closed-loop system from static analysis alerts to interpretable compliance reports;
Pricing Online LLM Services with Data-Calibrated Stackelberg Routing Game
Zhendong Guo (Southeast University), Jiahui Jin (Southeast University)
OptimizationData-Centric LearningTransformerTabularFinance Related
🎯 What it does: Proposes the PriLLM framework for real-time dynamic pricing of a single service provider in an LLM routing platform, modeling user-provider interactions based on Stackelberg game theory;
Primary Visual Cortex Inspired Point Cloud Analysis Framework
Jisheng Dang (Lanzhou University), Jizhao Liu (Lanzhou Jiaotong University)
ClassificationConvolutional Neural NetworkSpiking Neural NetworkPoint Cloud
🎯 What it does: This paper proposes a branching continuous coupled neural network (DC-CCNN) based on the primary visual cortex for point cloud analysis.
PRIME: Planning and Retrieval-Integrated Memory for Enhanced Reasoning
Hieu Tran (VA Bedford Health Care), Hong Yu (University of Massachusetts Amherst)
TransformerLarge Language ModelAgentic AITextBiomedical DataBenchmarkRetrieval-Augmented GenerationChain-of-Thought
🎯 What it does: Propose a multi-agent reasoning framework PRIME, combining fast intuition (System 1) with deep reasoning (System 2), using self-reflection to decide whether to activate slow reasoning;
Principled Analysis of Deep Reinforcement Learning Evaluation and Design Paradigms
Ezgi Korkmaz
Reinforcement LearningVideoBenchmark
🎯 What it does: This paper systematically analyzes the implicit assumptions in the evaluation and design paradigms of deep reinforcement learning through theoretical proofs and large-scale experiments, revealing the root cause of non-monotonic performance rankings between low-sample and high-sample regimes.
Prior Refinement Is Better: Diffusion-Driven Graph Harmonization for Federated Graph Learning
Shuman Zhuang (Fuzhou University), Hong-Ning Dai (Hong Kong Baptist University)
Federated LearningDiffusion modelAuto EncoderContrastive LearningGraph
🎯 What it does: Proposes the FedGH (Federated Graph Harmonization) framework, which pre-refines graph data through a graph diffusion model before federated graph learning to reduce semantic and topological heterogeneity among clients.
PriorDrive: Enhancing Online HD Mapping with Unified Vector Priors
Shuang Zeng (Xi'an Jiaotong University), Xing Wei (Alibaba Group)
Autonomous DrivingRepresentation LearningTransformer
🎯 What it does: Improve the completeness and accuracy of online high-definition map construction by unifying encoding and fusing multiple vectorized prior maps (e.g., OSM standard maps, legacy HD maps, and historical local prediction maps), enabling high-quality predictions even under occlusions, adverse weather, or remote areas.
Priority-Based Graph-Enhanced Reinforcement Learning for Robust Analog Circuit Optimization
Jintao Li, Shui Yu (University of Electronic Science and Technology of China)
OptimizationGraph Neural NetworkReinforcement LearningContrastive LearningGraphPhysics Related
🎯 What it does: Proposes a priority graph enhanced reinforcement learning framework, employing fuzzy logic to convert multi-objective rewards into priority signals, and utilizing a PVT variant graph to compress the high-dimensional target space, achieving robust analog circuit optimization.
PriorRG: Prior-Guided Contrastive Pre-training and Coarse-to-Fine Decoding for Chest X-ray Report Generation
Kang Liu (Xidian University), Qiguang Miao (Xidian University)
GenerationTransformerLarge Language ModelVision Language ModelContrastive LearningImageTextMultimodalityBiomedical DataRetrieval-Augmented Generation
🎯 What it does: This paper proposes the PriorRG framework, which utilizes patients' prior imaging, clinical indicators, and medical history information to generate chest X-ray reports that align better with clinical workflows.
PRISM: Privacy-Aware Routing for Adaptive Cloud–Edge LLM Inference via Semantic Sketch Collaboration
Junfei Zhan (University of Pennsylvania), Tengjiao He (University of Hong Kong)
Safty and PrivacyComputational EfficiencyTransformerLarge Language ModelText
🎯 What it does: Propose the PRISM framework to achieve cloud-edge collaborative LLM inference, utilizing dynamic routing, two-layer local differential privacy, and semantic sketch collaboration to enhance privacy and efficiency.
Privacy Auditing of Multi-Domain Graph Pre-Trained Model Under Membership Inference Attacks
Jiayi Luo (Beihang University), Jianxin Li (Guangxi Normal University)
Safty and PrivacyAdversarial AttackGraph Neural NetworkGraph
🎯 What it does: Propose the MGP-MIA framework for membership inference attacks on multi-domain graph pre-training models, enhancing the ability to detect privacy leakage
Privacy Leaks by Adversaries: Adversarial Iterations for Membership Inference Attack
Jing Xue (Xi'an Jiaotong University), Guang Dai (Fudan University)
Safty and PrivacyAdversarial AttackConvolutional Neural NetworkImage
🎯 What it does: Proposes a Membership Inference Attack (IMIA) based on the number of iterations required to generate adversarial samples, determining whether a sample is a member of the model's training set by measuring the number of iterations needed to generate adversarial samples.
Privacy on the Fly: A Predictive Adversarial Transformation Network for Mobile Sensor Data
Tianle Song (Xi'an Jiaotong University), Chao Shen (Xi'an Jiaotong University)
Safty and PrivacyComputational EfficiencyRecurrent Neural NetworkGenerative Adversarial NetworkTime Series
🎯 What it does: Proposes the Predictive Adversarial Transformation Network (PATN), a real-time privacy protection framework that predicts future adversarial perturbations using historical sensor signals;
Privacy Preserving In-Context-Learning Framework for Large Language Models
Bishnu Bhusal (University of Missouri), Susmit Jha (SRI International)
Data SynthesisSafty and PrivacyTransformerLarge Language ModelText
🎯 What it does: Proposed a differential privacy-based private prediction framework that generates high-quality synthetic text without fine-tuning large language models, and applies it to in-context learning (ICL).
Privacy-protected Retrieval-Augmented Generation for Knowledge Graph Question Answering
Yunfeng Ning (Wuhan University), Tieyun Qian (Wuhan University)
GenerationRetrievalSafty and PrivacyTransformerLarge Language ModelGraphRetrieval-Augmented GenerationChain-of-Thought
🎯 What it does: Proposes the ARoG framework to address retrieval challenges caused by entity anonymization in KGQA, achieving a retrieval-augmented generation (RAG) system under privacy-preserving conditions.
Private Frequency Estimation via Residue Number Systems
Héber Hwang Arcolezi (Inria)
Computational EfficiencyTabular
🎯 What it does: Propose a local differential privacy frequency estimation protocol based on the residue number system—Modular Subset Selection (MSS)—which achieves efficient frequency estimation with a single message.
PrivSV: Differentially Private Steering Vector for Large Language Models
Haocheng Yang (Beijing University of Posts and Telecommunications), Sen Su (Beijing University of Posts and Telecommunications)
OptimizationSafty and PrivacyConvolutional Neural NetworkLarge Language ModelText
🎯 What it does: Proposes a generic differential privacy method, PrivSV, compatible with any Steering Vector (SV) construction paradigm, aiming to strictly protect sensitive information contained in SV while maintaining efficient LLM control capabilities.
Proactive Constrained Policy Optimization with Preemptive Penalty
Ning Yang (Institute of Automation Chinese Academy of Sciences), Jun Wang (Microsoft Research)
OptimizationReinforcement Learning
🎯 What it does: This paper proposes Proactive Constrained Policy Optimization (PCPO), achieving safe reinforcement learning through proactive penalties and constraint-aware intrinsic rewards.
ProAR: Probabilistic Autoregressive Modeling for Molecular Dynamics
Kaiwen Cheng (Peking University), Yonghong Tian (Peking University)
Drug DiscoveryProtein Structure PredictionGraph Neural NetworkTransformerScore-based ModelFlow-based ModelTime SeriesSequentialBiomedical Data
🎯 What it does: Developed a probabilistic autoregressive framework, ProAR, which uses a dual network for frame-by-frame generation and interpolation of molecular dynamics trajectories;
Probabilistic Deformation Consistency for Unsupervised Shape Matching
Yifan Xia (Wuhan University), Jiayi Ma (Wuhan University)
RecognitionPose EstimationDiffusion modelPoint CloudMeshBenchmark
🎯 What it does: Proposed an unsupervised shape matching framework called PDCMatch, based on a probabilistic deformation consistency model in the spectral domain, which improves point correspondence by jointly estimating deformation and correspondence probabilities.
Probabilistic Hash Embeddings for Online Learning of Categorical Features
Aodong Li (Amazon Web Services), Balakrishnan Murali Narayanaswamy (Amazon Web Services)
ClassificationRecommendation SystemTabularTime Series
🎯 What it does: This paper proposes a probabilistic hashing embedding (PHE) model to handle scalable and dynamically changing categorical feature vocabularies in a streaming online learning environment.
Probabilistic Hierarchical Goal Network Planning with UCT
David H. Chan (University of Maryland), Dana S. Nau (U.S. Naval Research Laboratory)
OptimizationReinforcement LearningBenchmark
🎯 What it does: Proposed a probabilistic hierarchical goal network (Probabilistic HGN) planning framework and implemented two UCT-based solving algorithms.
Probabilistic Safety Verification of Neural Policies via Predicate Abstraction
Marcel Vinzent (Saarland University), Jörg Hoffmann (Saarland University)
Safty and PrivacyReinforcement LearningBenchmark
🎯 What it does: This paper proposes a probabilistic safety verification method for neural network control strategies in the presence of probabilistic transitions and non-determinism.
Probabilities Are All You Need: A Probability-Only Approach to Uncertainty Estimation in Large Language Models
Manh Nguyen (Deakin University), Hung Le (Deakin University)
Explainability and InterpretabilityTransformerLarge Language ModelText
🎯 What it does: Studied uncertainty estimation in LLMs, proposing a training-free estimation method called PRO based on top-K probabilities.
Probability Distribution Alignment and Low-Rank Weight Decomposition for Source-Free Domain Adaptive Brain Decoding
Ganxi Xu (Jinan University), Jia Zhang (Jinan University)
Domain AdaptationComputational EfficiencyVision Language ModelDiffusion modelImageBiomedical DataMagnetic Resonance Imaging
🎯 What it does: In brain decoding tasks, a framework based on source-free domain adaptation (SFDA) is proposed, addressing inter-subject differences and cross-modal alignment while reducing parameter count through low-rank weight decomposition.
ProBench: Benchmarking GUI Agents with Accurate Process Information
Leyang Yang (Zhejiang University), Yong Li (Ant Group)
Large Language ModelAgentic AITextBenchmark
🎯 What it does: Propose ProBench, a mobile benchmark containing over 200 multi-step GUI tasks covering state-related tasks and process-related tasks, and design an automatic evaluation pipeline.
Probing EFX via PMMS: (Non-)Existence Results in Discrete Fair Division
Jaroslaw Byrka, Tomasz Ponitka (Tel Aviv University)
Optimization
🎯 What it does: This paper investigates the existence of two important fairness concepts in discrete fair division—EFX and PMMS. It proves that PMMS may not exist under the setting of three agents and provides constructive proofs for the existence of PMMS or EFX in three special cases (personalized bi-valued, bi-valued MMS-feasible, and dual demand). Additionally, it provides polynomial-time algorithms.
Probing Preference Representations: A Multi-Dimensional Evaluation and Analysis Method for Reward Models
Chenglong Wang (Northeastern University), Tong Xiao (Northeastern University)
Explainability and InterpretabilityRepresentation LearningReinforcement Learning from Human FeedbackTransformerLarge Language ModelReinforcement LearningTextBenchmark
🎯 What it does: By constructing a six-dimensional MRMBench, the preference representation of reward models is evaluated using probe methods, and interpretability is enhanced through inference-time probe analysis.
Probing Semantic Insensitivity for Inference-Time Backdoor Defense in Multimodal Large Language Model
Xuankun Rong (Wuhan University), Mang Ye (Wuhan University)
Anomaly DetectionTransformerLarge Language ModelPrompt EngineeringVision Language ModelMultimodality
🎯 What it does: Proposes a reasoning-time anti-backdoor detection framework ToT based on text semantic perturbation, used to identify visual backdoors in multi-modal large language models;
ProbLog4Fairness: A Neurosymbolic Approach to Modeling and Mitigating Bias
Rik Adriaensen (KU Leuven), Maarten Buyl (Ghent University)
Explainability and InterpretabilityImageTabular
🎯 What it does: Propose the ProbLog4Fairness framework, modeling bias as a ProbLog program and integrating DeepProbLog for bias mitigation
ProCache: Constraint-Aware Feature Caching with Selective Computation for Diffusion Transformer Acceleration
Fanpu Cao (South China University Of Technology), Wei Luo (South China University Of Technology)
GenerationComputational EfficiencyTransformerDiffusion modelImageTextMultimodalityBenchmark
🎯 What it does: Proposes ProCache, a training-agnostic, dynamic feature caching framework to accelerate the inference of Diffusion Transformers (DiT);
ProCAST: A Projection Framework for Coupled Aggregation Constrained Multivariate Time Series Forecasting
Jiaqi Xue (Shanghai Jiao Tong University), Guihai Chen (Shanghai Jiao Tong University)
OptimizationTime Series
🎯 What it does: ProCAST proposes a projection framework that maps unconstrained multivariate time series predictions into feasible domains satisfying coupled aggregation constraints through orthogonal or oblique projections, ensuring prediction feasibility.
ProCrop: Learning Aesthetic Image Cropping from Professional Compositions
Ke Zhang, Luming Liang (Microsoft)
GenerationData SynthesisRetrievalTransformerLarge Language ModelVision Language ModelDiffusion modelImageMultimodalityRetrieval-Augmented Generation
🎯 What it does: Propose ProCrop, a retrieval-driven aesthetic image cropping method, and construct a large-scale weakly supervised cropping dataset with 242K entries.
ProFuser: Progressive Fusion of Large Language Models
Tianyuan Shi (Sun Yat-sen University), Wu Kai (Alibaba Group)
Knowledge DistillationTransformerLarge Language ModelSupervised Fine-TuningText
🎯 What it does: Proposed the ProFuser method, which progressively fuses multiple large language models using a dual evaluation mechanism combining the minimum cross-entropy of training modes and reward model voting in inference modes.
ProGMLP: A Progressive Framework for GNN-to-MLP Knowledge Distillation with Efficient Trade-offs
Weigang Lu (Hong Kong University of Science and Technology), Dapeng Tao (JD Explore Academy)
Computational EfficiencyKnowledge DistillationGraph Neural NetworkGraph
🎯 What it does: Designed a GNN-to-MLP knowledge distillation framework called ProGMLP, which adjusts inference cost and accuracy on demand through progressively trained multi-layer MLP students to achieve flexible inference.
ProgRAG: Hallucination-Resistant Progressive Retrieval and Reasoning over Knowledge Graphs
Minbae Park (Hanyang University), Hyunjoon Kim (Seoul National University)
RetrievalGraph Neural NetworkTransformerLarge Language ModelPrompt EngineeringGraphRetrieval-Augmented GenerationChain-of-Thought
🎯 What it does: Propose the ProgRAG framework, which decomposes complex knowledge graph question answering problems into sub-questions, iteratively retrieves information, and constructs optimal context through LLM pruning and prefix enumeration re-ranking;
Progressive Multi-modal Knowledge Distillation for Multi-spectral Object Re-identification
Aihua Zheng (Anhui University), Jin Tang (Anhui University)
RetrievalKnowledge DistillationTransformerMultimodality
🎯 What it does: This paper proposes a multi-stage multi-modal knowledge distillation framework aimed at enhancing the performance of multispectral object re-identification.
ProLoG: Hybrid Prompt and LoRA Based Adaptation of Vision-Language Models for OOD Generalization
Jungwuk Park (Korea Advanced Institute of Science and Technology), Jaekyun Moon (Korea Advanced Institute of Science and Technology)
ClassificationDomain AdaptationRepresentation LearningTransformerPrompt EngineeringVision Language ModelMultimodality
🎯 What it does: This paper proposes a hybrid prompting and LoRA adaptation method called ProLoG, which combines prompt tuning, LoRA low-rank matrices, and augmented regularization. During inference, it dynamically selects prompts and LoRA based on task similarity to enhance the generalization ability of vision-language models in out-of-distribution (OOD) scenarios.
PROMISE: Prompt-Attentive Hierarchical Contrastive Learning for Robust Cross-Modal Representation with Missing Modalities
Jiajun Chen (Beijing University of Posts and Telecommunications), Yi Zhong (Beijing Institute of Technology)
Representation LearningTransformerPrompt EngineeringContrastive LearningMultimodality
🎯 What it does: Propose the PROMISE framework, combining multi-modal prompt learning with hierarchical contrastive learning to achieve robust cross-modal representation learning in missing modality scenarios.