Research

Spatiotemporal Data Mining focuses on extracting knowledge and patterns from massive datasets with temporal and spatial dimensions, particularly large-scale trajectory data from moving objects. It involves developing machine learning-based modeling frameworks and computational paradigms to tackle real-world challenges in downstream spatiotemporal applications.

Graph Neural Networks (GNNs) specialize in learning effective representation learning for large-scale complex graph-structured data, with particular emphasis on massive multi-layer heterogeneous graph networks. This research domain focuses on developing scalable GNN architectures tailored for complex graph networks with billions of nodes/edges, while enhancing performance across diverse graph mining tasks such as recommendation systems, anomaly detection, and beyond.

Graph Data Mining focuses on discovering frequent and meaningful subgraph patterns (e.g., network motifs) from large-scale temporal graphs. It involves designing highly scalable and parallelizable algorithms to enable efficient mining and analysis of massive temporal graphs on modern multi-core computing architectures.

Intelligent Transportation Systems aim to enhance urban mobility by leveraging large-scale traffic observation data and advanced machine learning techniques. This research direction focuses on developing data-driven predictive models for accurate traffic flow estimation, including vehicle and pedestrian dynamics, to optimize existing urban transportation infrastructure.