Cross Domain Reasoning Based on Graph Deep Learning

Cross domain reasoning has garnered significant attention as a cutting-edge yet challenging research area with numerous practical applications. On the other hand, representing pairwise information via Graph Structure is gaining popularity due to combining ideas from mathematics, physics, biology, computer science, statistics, and many other areas. Cross domain reasoning of graph deep learning, brings the challenges such as heterogeneous knowledge integration, implicit knowledge mining restricted by various cognitive theory-driven constraints, spatio-temporal relation utilization, and problems of tractability and optimization. Therefore, the general goal of this research is to develop the graph deep learning models for cross domain reasoning. To solve the above problems, there are two main aims to be achieved: (1) multi-task learning and (2) task relation reasoning. To achieve these two goals, we proposed four unique models (DETECTIVE, DynAttGraph2Seq, RAPTA, DeepGAR) which have the capabilities of capturing heterogeneous spatiotemporal relationships, dealing with incomplete data, integrating existing rules, and transferring knowledge. Extensive experiments were conducted on synthetic and real-world datasets to demonstrate the effectiveness of the proposed models for real-world applications including but not limited to static timing analysis, healthcare, spatiotemporal event forecasting, analogical reasoning, etc.

Author

• Chowdhury, Tanmoy

• cross-domain
• deep learning
• GNN
• multi-task learning
• reasoning
• spatiotemporal events

Identifier

hdl:1920/13950

OAI identifier oai:identifier

oai:MARS:1920/13950