top of page
Time-Varying Interaction Analysis
Exploring time-directed interactions in mutivariate data has been challenged among data sciences for decades. Adaptive Directed information (ADI) is a useful tool and has been designed to discover complex dependencies between entities. In this project we develop novel time-based version of ADI to allocate an importance window to enteties as prior additional information. These techniques to interaction estimation are applied in various domains such as recurrent neural network efficient training, agents causality, interaction detection in transportation of crowded cities, etc.
bottom of page