Abstract:We have witnessed a rapid growth over past decades in sensor data mining (SDM), which aims at extracting valuable information automatically from large repositories of moving activity data. One of the significant SDM tasks is identifying humans through their transit modes using a variety of user-tracking systems. However, to the best of our knowledge, distinguishing traces of users and understanding their behaviors are difficult tasks in most real-life cases for the following reasons: 1) activity data containing both temporal and spatial contexts are of high order and sparse; 2) living patterns are not as regular as expected, and the route choice uncertainties due to their vagueness and randomness; 3) owing to the complexity and sparseness of urban travel methods, although some deep learning-based models can produce relatively good classification results, they can still be improved by combining external information. To address these challenges, we propose a novel scenario-based deep learning method which is based on the assumption that people visit places with explicit purposes (e.g., to go to work or visit a park). We first represent semantic patterns from daily life and create various scenarios and utilize an attention neural network to embed points of trajectories by considering both semantic and geographical information. Then, we construct a Semantic Structure Neural Network (S2N2) framework to perform the end-end classification. Our S2N2 model is applied to an interesting yet challenging topic: distinguishing suspect transit behavior on a real-life data set collected by mobile devices. Although the problem is not entirely solved, the extensive evaluation presented here demonstrates that our model outperforms conventional classification methods, anomaly detection methods, and state-of-the-art sequential deep learning models, especially when trajectory semantic vectors are incorporated. We also provide statistical analysis and intuitive explanations to help interpret the characteristics of user mobility.

Inferring Ongoing Human Activities Based on Recurrent Self-Organizing Map Trajectory.

Human activity prediction by mapping grouplets to recurrent Self-Organizing Map.

Deep Behavior Tracing with Multi-level Temporality Preserved Embedding

Human Activity Recognition based on Dynamic Spatio-Temporal Relations

Deep Dilation on Multimodality Time Series for Human Activity Recognition.

Activity Recognition Using Dense Long-Duration Trajectories

Discovery and Recognition of Emerging Human Activities Using a Hierarchical Mixture of Directional Statistical Models.

A Hierarchical Spatio-Temporal Model for Human Activity Recognition.

Salient Pairwise Spatio-Temporal Interest Points for Real-Time Activity Recognition.

SOM Based Activity Learning for Visual Surveillance System

Human Activities Prediction by Learning Combinatorial Sparse Representations.

S2N2: An Interpretive Semantic Structure Attention Neural Network for Trajectory Classification

Continuous Human Activity Recognition through Parallelism LSTM with Multi-Frequency Spectrograms

Predicting Complex Activities from Ongoing Multivariate Time Series

Poster: Attention-based Spatio-Temporal Model for HAR Using Multivariate Time Series.

Temporal Reasoning Graph For Activity Recognition

Structured Feature-Graph Model for Human Activity Recognition

SOM based activity learning for visual topological system

Regions are Who Walk Them: a Large Pre-trained Spatiotemporal Model Based on Human Mobility for Ubiquitous Urban Sensing

Spatial-temporal activity interactions detection in video survalliance

Learning spatio-temporal co-occurrence correlograms for efficient human action classification