Colloquium Presentation By: Shaon Bhatta Shuvo

SCHOOL OF COMPUTER SCIENCE – Colloquium Series

The School of Computer Science at the University of Windsor is pleased to present…

Title: Deep Reinforcement Learning-Based Optimization for Complex System Analysis in Computational Epidemiology

Colloquium Presentation By: Shaon Bhatta Shuvo

Date: Friday, 05 Apr 2024

Time: 11:00 am - 12:00 pm

Location: Erie Hall, room 3123

Abstract:

Complex systems, characterized by intricate interconnections and emergent behaviours, challenge the conventional paradigms of analysis and optimization. These systems permeate various domains, including biological ecosystems, social networks, and technological infrastructures, necessitating advanced computational methods for effective management. In the field of computational epidemiology, this study presents a decision support system employing Deep Reinforcement Learning (DRL) to optimize the complex dynamics of infectious disease spread, aimed at enhancing pandemic preparedness and response strategies. By integrating Agent-Based Modeling (ABM) with an extended SEIHRD (Susceptible, Exposed, Infectious, Hospitalized, Recovered, and Dead) framework through N-step Deep Q Reinforcement Learning (N-Step DQRL), the model enhances non-pharmaceutical interventions like lockdown policies. This hybrid, data-driven approach provides dynamic decision-making capabilities, essential for effective pandemic response in the absence of immediate vaccines. Tested against recent COVID-19 data, the model demonstrates its potential to improve pandemic preparedness and response, offering a scalable and detailed tool for policymakers to manage future health crises efficiently.

Keywords: agent-based modelling, compartmental model, deep learning, reinforcement learning, optimization

Biography:

Shaon is a Ph.D. Candidate and Sessional Instructor in the School of Computer Science at the University of Windsor. He also holds a part-time position as a Deep Learning Researcher at Magna International. His research spans multiple Artificial Intelligence domains, including Machine Learning, Deep Learning, and Agent-Based Modeling, with experience as a Research Assistant on various CIHR and NSERC-funded projects. Shaon has published in journals and presented at top-tier conferences, such as KDD, receiving the best paper award at SSDBM’20. Awarded multiple scholarships, including the OGS, he has over five years of university-level teaching experience across various universities internationally and has earned excellent teaching awards.