SCAI Spring ’22 Seminar Series Invited Speaker Manish Bansal presents algorithms for network vulnerability analysis and camera-based surveillance in this talk hosted by Assistant Professor Adolfo Escobedo.
Exact Algorithms for Distributionally Risk-Receptive Programs and Camera View-Frame Placement Problems: Applications in Homeland Security
Presented by Manish Bansal
For many security applications, it is critical to make strategic long-term planning decisions with uncertain input data parameters that also allow adjustments based on risk-appetite of a decisi0n-maker.
In the first half of this talk, Manish Bansal considers a distributional risk-receptive network interdiction problem, or DRR-NIP, where a leader maximizes a follower’s minimal expected objective value for the best-case probability distribution belonging to a given set of distributions, referred to as an ambiguity set. The DRR-NIP is applicable for network vulnerability analysis where a network user seeks to identify vulnerabilities in the network against potential disruptions by an adversary who is receptive to risk for improving the expected objective values. Bansal presents exact and approximation algorithms for solving DRR-NIP with a general ambiguity set. He also provides conditions for which these approaches are finitely convergent, along with the results of his team’s extensive computational experiments.
In the second half of the talk, he introduces a combinatorial optimization problem pertinent to network-based telerobotic cameras that enable decision-makers to interact with a remote physical environment using shared resources. Specifically, we consider a system of p networked robotic cameras that receives rectangular subregions as requests from multiple users for monitoring. Each subregion or request has an associated reward rate that depends on the importance level associated with monitoring that subregion. Bansal’s team’s goal is to select the best view frame (pan, tilt and zoom) for the cameras with discrete or continuous resolutions to maximize the total reward from the captured parts of the requested subregions. His team develops exact and approximation algorithms for solving this NP-hard problem. These optimal or near-optimal solutions provide information to decision-makers to conduct surveillance and reconnaissance in environments where it is tedious for humans to collect information. He also presents the results of his team’s computational experiments conducted to evaluate the performance of these algorithms.
About the speaker
Manish Bansal is an assistant professor with the Grado Department of Industrial and Systems Engineering at Virginia Tech. He completed his bachelor’s degree in electrical engineering from the National Institute of Technology in India and a master’s degree with a thesis and a doctoral degree from the Department of Industrial and Systems Engineering at Texas A&M University.
Prior to joining Virginia Tech, he was a postdoctoral fellow in the Department of Industrial Engineering and Management Sciences at Northwestern University. He has served as president and vice president of INFORMS Junior Faculty Interest Group from 2020 through 2022. He is serving as secretary of the Engineering Faculty Organization at Virginia Tech.
Bansal’s research is focused on the theory of mixed-integer programming, stochastic and distributional robust optimization and location science, along with their applications in homeland security, logistics and telerobotics. Currently, his research team has four doctoral students and has received multiple grants from the National Science Foundation and the U.S. Department of Defense.