Prof. Jiming LIU

1. Demystifying deep learning: An information-theoretic framework

Deep learning has achieved incredible success in various challenging predictive spatiotemporal analytics (PSTA) tasks. However, given a specific PSTA task, how to appropriately determine the desired architecture of a deep learning model remains a mystery. To demystify the power of deep learning for PSTA in a theoretically sound and explainable way, we provide a comprehensive framework for deep learning model design and information-theoretic analysis. In so doing, we tackle an important open question in deep learning, that is, how to theoretically determine the necessary and sufficient configurations of a deep learning model, with respect to given learning datasets.

2. How COVID-19 transmits: A social-contact characterization

COVID-19 has spread to 6 continents. Gaining a deeper understanding of what may have happened could potentially inform mitigation strategies in disease-emerging countries. This work addresses how COVID-19 spreads among different populations and the corresponding impacts on disease control, e.g., with/without interventions. We approach this question by examining age-specific social contact-based transmissions. The results show that the social contact-based analysis can readily explain the underlying disease transmission patterns and associated potential risks. This work has uncovered the ins and outs of COVID-19 outbreaks, which could have practical implications to mitigation strategy planning in various countries and regions. (Please view the related output here)