AAAI 2025 Papers — Page 31
AAAI Conference on Artificial Intelligence · 3028 papers
WiFi CSI Based Temporal Activity Detection via Dual Pyramid Network
Zhendong Liu (University of Electronic Science and Technology of China), Ce Zhu (Sichuan University)
RecognitionAnomaly DetectionConvolutional Neural NetworkTransformerTime Series
🎯 What it does: A dual-pyramid network DPWiT is proposed for time activity detection of untrimmed long-term WiFi CSI signals.
WildFake: A Large-Scale and Hierarchical Dataset for AI-Generated Images Detection
Yan Hong (Ant Group), Jianfu Zhang (Shanghai Jiao Tong University)
GenerationTransformerDiffusion modelGenerative Adversarial NetworkImageBenchmark
🎯 What it does: A large-scale AI-generated image detection dataset, WildFake, has been constructed and made publicly available, providing generated images across multiple categories, architectures, weights, times, and versions, and evaluating the generalization and robustness of detectors based on this dataset.
Wills Aligner: Multi-Subject Collaborative Brain Visual Decoding
Guangyin Bao (Tongji University), Duoqian Miao (Macquarie University)
ClassificationRetrievalMeta LearningMixture of ExpertsContrastive LearningImageMagnetic Resonance Imaging
🎯 What it does: A multi-subject collaborative brain visual decoding framework called Wills Aligner is proposed, which can simultaneously process fMRI data from different subjects and decode visual information.
Windowed MAPF with Completeness Guarantees
Rishi Veerapaneni (Carnegie Mellon University), Maxim Likhachev (Carnegie Mellon University)
OptimizationBenchmark
🎯 What it does: A complete framework for window-based multi-agent pathfinding (WinC-MAPF) is proposed, and within this framework, a single-step conflict-based search (SS-CBS) is designed, which can optimally plan at each step while ensuring theoretical completeness.
Witty: An Efficient Solver for Computing Minimum-Size Decision Trees
Luca Pascal Staus (Friedrich Schiller University Jena), Manuel Sorge (TU Wien)
ClassificationOptimizationComputational EfficiencyTabularBenchmark
🎯 What it does: A minimum size decision tree solver called Witty based on witness-tree is proposed and implemented, with several heuristic improvements added on top of it.
World Knowledge-Enhanced Reasoning Using Instruction-Guided Interactor in Autonomous Driving
Mingliang Zhai (Beijing Institute of Technology), Yunde Jia (Beijing Institute of Technology)
Autonomous DrivingTransformerLarge Language ModelSupervised Fine-TuningVideoTextMultimodality
🎯 What it does: This paper proposes an instruction-guided interactive module (Interactor) that achieves reasoning and decision-making of multimodal large language models in perception-constrained scenarios by pre-fusing multi-view videos and world knowledge.
WorldAPIs: The World Is Worth How Many APIs? A Thought Experiment
Jiefu Ou (Johns Hopkins University), Daniel Khashabi (Johns Hopkins University)
AI Code AssistantTransformerLarge Language ModelPrompt EngineeringText
🎯 What it does: This paper proposes a thought experiment called WORLDAPIS based on wikiHow tutorials, which uses LLM to gradually generate Python programs to induce and expand a collection of executable APIs, thereby estimating the available primitive action space in the real world.
WPMixer: Efficient Multi-Resolution Mixing for Long-Term Time Series Forecasting
Md Mahmuddun Nabi Murad (University of South Florida), Yasin Yilmaz (University of South Florida)
Computational EfficiencyHyperparameter SearchReinforcement LearningTime SeriesSequential
🎯 What it does: A long sequence time series prediction model called WPMixer is proposed, which is based on multi-scale wavelet decomposition and MLP Mixer.
WST: Wavelet-Based Multi-scale Tuning for Visual Transfer Learning
Jia Zeng (Jilin University), Kangping Wang (Jilin University)
ClassificationDomain AdaptationTransformerSupervised Fine-TuningImage
🎯 What it does: A parameter-efficient fine-tuning method WST based on small-scale partitioning and wavelet transform is designed;
XCOT: Cross-lingual Instruction Tuning for Cross-lingual Chain-of-Thought Reasoning
Linzheng Chai (Beihang University), Zhoujun Li (Beihang University)
Knowledge DistillationTransformerLarge Language ModelSupervised Fine-TuningTextChain-of-Thought
🎯 What it does: Enhance the reasoning ability of low-resource languages through cross-lingual instruction fine-tuning, random online chain-of-thought reasoning, and cross-lingual distillation.
xPatch: Dual-Stream Time Series Forecasting with Exponential Seasonal-Trend Decomposition
Artyom Stitsyuk (Korea Advanced Institute of Science and Technology), Jaesik Choi (Korea Advanced Institute of Science and Technology)
OptimizationConvolutional Neural NetworkRecurrent Neural NetworkTime Series
🎯 What it does: A dual-stream (MLP+CNN) structure called xPatch is proposed, which achieves long sequence time series prediction through exponential moving average decomposition combined with channel independence and slicing techniques.
You Should Learn to Stop Denoising on Point Clouds in Advance
Chuchen Guo (China University of Geosciences), Ying He (Nanyang Technological University)
RestorationConvolutional Neural NetworkPoint Cloud
🎯 What it does: The Adaptive Stop Denoising Network (ASDN) is proposed, which prevents over-smoothing by adaptively stopping the denoising of cleaned points, thereby improving the quality of point cloud denoising.
Yuan: Yielding Unblemished Aesthetics Through a Unified Network for Visual Imperfections Removal in Generated Images
Zhenyu Yu (Universiti Malaya), Chee Seng Chan (Universiti Malaya)
RestorationSegmentationGenerationSupervised Fine-TuningImage
🎯 What it does: Eliminate visual defects in text-generated images through automated segmentation and repair processes.
Zero-resource Hallucination Detection for Text Generation via Graph-based Contextual Knowledge Triples Modeling
Xinyue Fang (National University of Defense Technology), Dongsheng Li (National University of Defense Technology)
GenerationAnomaly DetectionGraph Neural NetworkLarge Language ModelTextGraph
🎯 What it does: This paper proposes a zero-resource hallucination detection method for long text generation, utilizing a graph structure to perform consistency comparisons on extracted knowledge triples, and strengthening detection through three reverse verification tasks.
Zero-Shot Conditioning of Score-Based Diffusion Models by Neuro-Symbolic Constraints
Davide Scassola (University of Trieste), Luca Bortolussi (University of Trieste)
GenerationData SynthesisDiffusion modelScore-based ModelImageTabularTime SeriesStochastic Differential Equation
🎯 What it does: A zero-shot conditional sampling method without training is proposed, utilizing a pre-trained unconditional score diffusion model and soft constraints to sample under arbitrary logical constraints.
Zero-shot Depth Completion via Test-time Alignment with Affine-invariant Depth Prior
Lee Hyoseok (POSTECH), Tae-Hyun Oh (POSTECH)
RestorationDepth EstimationDomain AdaptationDiffusion modelPoint Cloud
🎯 What it does: A zero-shot depth completion method based on a pre-trained deep diffusion model and test-time alignment is proposed.
Zero-Shot Image Captioning with Multi-type Entity Representations
Delong Zeng (Sun Yat-Sen University), Jiarui Ouyang (Sun Yat-Sen University)
GenerationRetrievalDomain AdaptationTransformerLarge Language ModelContrastive LearningImageTextRetrieval-Augmented Generation
🎯 What it does: This paper proposes a zero-shot image captioning method called MERCap, which utilizes multi-type entity representations for retrieval and generates descriptions through GPT-2.
Zero-Shot Learning for Materials Science Texts: Leveraging Duck Typing Principles
Xin Zhang (Wuhan University of Technology), Lin Li (Wuhan University of Technology)
TransformerLarge Language ModelPrompt EngineeringText
🎯 What it does: Proposes MatDuck, a zero-shot materials science text mining method based on the duck typing principle;
Zero-Shot Learning in Industrial Scenarios: New Large-Scale Benchmark, Challenges and Baseline
Zekai Zhang (Shandong University), Jinglin Zhang (Shandong University)
Object DetectionDomain AdaptationConvolutional Neural NetworkContrastive LearningImageMultimodalityBenchmark
🎯 What it does: Proposed the MMIO industrial defect open dataset and the RTVP zero-shot detection method based on expert guidance, sparse modeling, and cross-modal interaction.
Zero-Shot Low-Light Image Enhancement via Latent Diffusion Models
Yan Huang (South China University of Technology), Yong Xu (South China University of Technology)
RestorationDiffusion modelImage
🎯 What it does: A zero-shot low-light image enhancement framework is proposed, utilizing a pre-trained latent diffusion model to achieve low-light enhancement without the need for specialized training.
Zero-Shot Noise2Mean: Gap Minimization for Efficient Denoising from a Single Noisy Image
Duo Liu (Harbin Engineering University), Liguo Zhang (Harbin Engineering University)
RestorationConvolutional Neural NetworkImageMagnetic Resonance Imaging
🎯 What it does: A zero-shot image denoising method ZS-N2M is proposed, which can train and achieve high-quality denoising results using only a single noisy image;
Zero-Shot Scene Change Detection
Kyusik Cho (Yonsei University), Euntai Kim (Yonsei University)
Object TrackingSegmentationAnomaly DetectionImageVideo
🎯 What it does: This paper proposes a zero-shot scene change detection method that does not require training, utilizing the pre-trained segmentation model SAM and the tracking model DEVA to achieve change recognition across temporal images.
Zero-shot Video Moment Retrieval via Off-the-shelf Multimodal Large Language Models
Yifang Xu (Nanjing University), Sidan Du (Dalian University of Technology)
RetrievalTransformerLarge Language ModelVideoTextMultimodality
🎯 What it does: This paper proposes Moment-GPT, a zero-shot video moment retrieval (VMR) framework based entirely on a frozen multimodal large language model (MLLM), capable of directly locating semantic segments in unedited videos without any fine-tuning.
Zero-shot Video Restoration and Enhancement Using Pre-Trained Image Diffusion Model
Cong Cao (Tianjin University), Jingyu Yang (Lappeenranta-Lahti University of Technology)
RestorationSuper ResolutionDiffusion modelOptical FlowVideo
🎯 What it does: A zero-shot video restoration and enhancement framework is proposed, utilizing a pre-trained image diffusion model and incorporating spatiotemporal attention, temporal consistency guidance, spatial-temporal noise sharing, and early stopping sampling.
ZeroHAR: Sensor Context Augments Zero-Shot Wearable Action Recognition
Ranak Roy Chowdhury (University of California San Diego), Jingbo Shang (University of California San Diego)
ClassificationRecognitionTransformerLarge Language ModelContrastive LearningMultimodalityTime Series
🎯 What it does: The ZeroHAR framework is proposed, which enhances zero-shot wearable action recognition by utilizing sensor context information (sensor type, axis, body position).
ZeroMamba: Exploring Visual State Space Model for Zero-Shot Learning
Wenjin Hou (Zhejiang University), Yi Yang (Zhejiang University)
ClassificationRecognitionRepresentation LearningContrastive LearningImage
🎯 What it does: This paper proposes ZeroMamba, a zero-shot learning framework based on Vision Mamba, which integrates three main modules: semantic-aware local projection, global representation learning, and semantic fusion, enhancing visual-semantic interaction.
Zeroth-Order Methods for Nonconvex Stochastic Problems with Decision-Dependent Distributions
Yuya Hikima (University of Tokyo), Akiko Takeda (University of Tokyo)
OptimizationGaussian SplattingTabular
🎯 What it does: Two zeroth-order methods for decision-related stochastic non-convex optimization problems are proposed, along with convergence and sample complexity analysis.
ZoRI: Towards Discriminative Zero-Shot Remote Sensing Instance Segmentation
Shiqi Huang (Nanyang Technological University), Bihan Wen (Nanyang Technological University)
Object DetectionSegmentationTransformerPrompt EngineeringVision Language ModelImage
🎯 What it does: A framework for zero-shot remote sensing instance segmentation, ZoRI, has been developed, utilizing the CLIP model to achieve segmentation of unseen categories.