Talk Title: Robust Incident Forecasting and Response
Emergency response to spatial-temporal incidents like crimes and accidents is a major challenge in today’s world. This talk will focus on two problems in this context. In the first segment, we will look at how robust machine learning models can help combat crimes like illegal poaching. While there are a variety of incident prediction models available to identify spatial-temporal patterns in such incidents, they fail to take attacker evasion into account. Specifically, poachers can shift their spatial preferences for committing crimes by observing patrols. This talk will present a general framework that models the interaction between off-the-shelf machine learning algorithms and attackers as a Stackelberg game. We will look at how such a game model works and how it can be solved to design patrol strategies. In the second segment, the talk will focus on how creating decentralized multi-agent approaches can improve emergency response in urban areas.
Ayan Mukhopadhyay is a Post-Doctoral Research Fellow at the Stanford Intelligent Systems Lab at Stanford University, USA. His research interests include multi-agent systems, robust machine learning and decision-making under uncertainty. He was awarded the 2019 CARS post-doctoral fellowship by the Center of Automotive Research at Stanford (CARS). Before joining Stanford, he finished his PhD at Vanderbilt University’s Computational Economics Research Lab. His doctoral thesis was nominated for the Victor Lesser Distinguished Dissertation Award 2020. His work on urban emergency response management has been covered in the Government Technology Magazine, Financial Times, multiple global smart city summits, and received a best paper award at ICLR’s AI for Social Good Workshop.