会议介绍

AAAI的英文全称是 The Association for the Advance of Artificial Intelligence，中文意思是美国人工智能协会。

美国人工智能协会（American Association for Artificial Intelligence）是人工智能领域的主要学术组织之一。该协会主办的年会（AAAI, The National Conference on Artificial Intelligence）是一个主要的人工智能学术会议，被中国计算机协会推荐为A类会议。

今年 AAAI 2021 将于北京时间2021年2月9到12号于线上举行。本次大会共接收了 1692 篇论文，接收率约为21%，可谓收获颇丰。编者梳理与时间序列有关的研究，竟也高达66篇，可见产学两界对该领域的热情与应用前景。完整的接收论文列表可以访问原文获取。

本文梳理AAAI 2021有关时间序列领域的最新研究成果，供大家参考。

• 时间序列

• • 预测：4篇

  • 分类：4篇

  • 异常检测：3篇

  • 聚类：1篇

  • 补缺：1篇

• 时间关系抽取：3篇

• 时空网络

  • 分割：3篇

  • 预测：13篇

  • 检测：11篇

• 时间知识图谱：3篇

• 时间神经网络：5篇

• 顺序分析

• • 推荐：5篇

  • 决策：7篇

  • 搜索：3篇

时间序列

01

预测

时序预测是时间序列领域的经典问题之一。本次AAAI带来了包括：多维时序预测/分解/多步预测的研究

• Deep Switching Auto-Regressive Factorization: Application to Time Series Forecasting

• Dynamic Gaussian Mixture Based Deep Generative Model for Robust Forecasting on Sparse Multivariate Time Series

• Temporal Latent Autoencoder: A Method for Probabilistic Multivariate Time Series Forecasting

• Synergetic Learning of Heterogeneous Temporal Sequences for Multi-Horizon Probabilistic Forecasting

02

分类

时序分类是时间序列领域的经典问题之一。本次AAAI带来了包括：多维时序分类/标签增强 等研究

• Correlative Channel-Aware Fusion for Multi-View Time Series Classification

• Learnable Dynamic Temporal Pooling for Time Series Classification

• ShapeNet: A Shapelet-Neural Network Approach for Multivariate Time Series Classification

• Joint-Label Learning by Dual Augmentation for Time Series Classification

03

异常检测

时序异常检测在产业界应用很广。本次AAAI带来了包括：图网络/异常影响等研究

• Graph Neural Network-Based Anomaly Detection in Multivariate Time Series

• Time Series Anomaly Detection with Multiresolution Ensemble Decoding

• Outlier Impact Characterization for Time Series Data

04

聚类

本次AAAI带来了一篇关于不完整时序的聚类研究

• Learning Representations for Incomplete Time Series Clustering

05

补缺

本次AAAI带了一篇关于多维时间序列补缺的研究

• Generative Semi-Supervised Learning for Multivariate Time Series Imputation

时间关系抽取

关系抽取旨在分析不同时序之间相关性，涉及因果推导。本次AAAI带来了3篇研究

• Clinical Temporal Relation Extraction with Probabilistic Soft Logic Regularization and Global Inference

• Time Series Domain Adaptation via Sparse Associative Structure Alignment

• Transformer-Style Relational Reasoning with Dynamic Memory Updating for Temporal Network Modeling

时空网络

01

分割

时空分割在视频分段场景应用广泛。本次AAAI带来了3篇研究

• Spatiotemporal Graph Neural Network Based Mask Reconstruction for Video Object Segmentation

• Temporal Relational Modeling with Self-Supervision for Action Segmentation

• Temporal Segmentation of Fine-Gained Semantic Action: A Motion-Centered Figure Skating Dataset

02

预测

时空预测在交通预测，用户行为预测，多因子气象预测等领域应用广泛。本次AAAI带来了13篇研究

• Graph and Temporal Convolutional Network for Spatio-Temporal 3D Multi-Person Pose Estimation in Monocular Videos

• Robust Spatio-Temporal Purchase Prediction via Deep Meta Learning

• BSN++: Complementary Boundary Regressor with Scale-Balanced Relation Modeling for Temporal Action Proposal Generation

• Automatic Generation of Flexible Plans via Diverse Temporal Planning

• Pre-Training Context and Time Aware Location Embeddings from Spatial-Temporal Trajectories for User Next Location Prediction

• GSNet: Learning Spatial-Temporal Correlations from Geographical and Semantic Aspects for Traffic Accident Risk Forecasting

• Spatial-Temporal Fusion Graph Neural Networks for Traffic Flow Forecasting

• Temporal Pyramid Network for Pedestrian Trajectory Prediction with Multi-Supervision

• Joint Air Quality and Weather Prediction Based on Multi-Adversarial Spatiotemporal Networks

• CloudLSTM: A Recurrent Neural Model for Spatiotemporal Point-Cloud Stream Forecasting

• Clinical Risk Prediction with Temporal Probabilistic Asymmetric Multi-Task Learning

• Traffic Flow Forecasting with Spatial-Temporal Graph Diffusion Network

• FC-GAGA: Fully Connected Gated Graph Architecture for Spatio-Temporal Traffic Forecasting

03

检测

时空检测在动作识别，目标检测，事件检测等场景应用广泛。本次AAAI带来了11篇研究

• Spatio-Temporal Difference Descriptor for Skeleton-Based Action Recognition

• Spatial-Temporal Causal Inference for Partial Image-to-Video Adaptation

• Temporal ROI Align for Video Object Recognition

• Multi-Scale Spatial Temporal Graph Convolutional Network for Skeleton-Based Action Recognition

• Learning Precise Temporal Point Event Detection with Misaligned Labels

• Real-Time Tropical Cyclone Intensity Estimation by Handling Temporally Heterogeneous Satellite Data

• STELAR: Spatio-Temporal Tensor Factorization with Latent Epidemiological Regularization

• ACSNet: Action-Context Separation Network for Weakly Supervised Temporal Action Localization

• Weakly-Supervised Temporal Action Localization by Uncertainty Modeling

• Weakly Supervised Temporal Action Localization through Learning Explicit Subspaces for Action and Context

• A Hybrid Attention Mechanism for Weakly-Supervised Temporal Action Localization

时空知识图谱

时空知识图谱在原有知识图谱上，考虑时间先后的知识变迁。本次AAAI带来3篇研究

• ChronoR: Rotation Based Temporal Knowledge Graph Embedding

• Learning from History: Modeling Temporal Knowledge Graphs with Sequential CopyGeneration Networks

• Neural Latent Space Model for Dynamic Networks and Temporal Knowledge Graphs

时间神经网络

处理时间先后顺序（或序列顺序）的神经网络为时序应用研究提供基础。本次AAAI带来了5篇研究

• Bridging Towers of Multi-Task Learning with a Gating Mechanism for Aspect-Based Sentiment Analysis and Sequential Metaphor Identification

• C2F-FWN: Coarse-to-Fine Flow Warping Network for Spatial-Temporal Consistent Motion Transfer

• Inductive Graph Neural Networks for Spatiotemporal Kriging

• Temporal-Coded Deep Spiking Neural Network with Easy Training and Robust Performance

• Continuous-Time Attention for Sequential Learning

顺序分析

01

推荐

分析用户购买行为的先后顺序是推荐系统的一大方向。本次AAAI带来了5篇研究

• Cold-Start Sequential Recommendation via Meta Learner

• A User-Adaptive Layer Selection Framework for Very Deep Sequential Recommender Models

• Noninvasive Self-Attention for Side Information Fusion in Sequential Recommendation

• Dynamic Memory Based Attention Network for Sequential Recommendation

• Reinforcement Learning of Sequential Price Mechanisms

02

决策

与推荐类似，用户决策具有一定的先后顺序。本次AAAI带来了7篇研究

• Model-Free Online Learning in Unknown Sequential Decision Making Problems and Games

• Bandit Linear Optimization for Sequential Decision Making and Extensive-Form Games

• Sequential Attacks on Kalman Filter-Based Forward Collision Warning Systems

• Apparently Irrational Choice as Optimal Sequential Decision Making

• Ethically Compliant Sequential Decision Making

• Hindsight and Sequential Rationality of Correlated Play

• Stock Selection via Spatiotemporal Hypergraph Attention Network: A Learning to Rank Approach

03

搜索

如何预估用户的行为从而提高搜索的效率？本次AAAI带来了3篇研究

• When Hashing Met Matching: Efficient Spatio-Temporal Search for Ridesharing

• Synthesis of Search Heuristics for Temporal Planning via Reinforcement Learning

• Sequential End-to-End Network for Efficient Person Search