AI for Networked Decision Making

Abstract

Today's world is marked by networked systems, such as communication networks, civil infrastructure, transportation and supply chains, social media, and health and education among communities. Decision making within a network often requires broader context of interactions with other entities. My most recent research revolves around the application of decision making algorithms, namely reinforcement learning, to automate planning and control of embodied and networked systems. I present results from applied reinforcement learning to the control of high speed optical communication systems in highly dynamic and non-stationary environments, as well as non-terrestrial networks, demonstrating the power of combining graph neural networks with deep reinforcement learning (GNN-DRL) to achieve significant performance gain and runtime improvements over other algorithms. Most importantly, graph-theoretic problems that can be solved via GNN-DRL can be applied to a variety of other important optimization problems in various networked systems. As I seek to lean my research direction toward achieving equitable outcomes in underserved communities, I will also discuss some helpful frameworks for achieving equitable resource allocation in networked systems, and describe areas in which AI for networked decision making can benefit public health and education.

Biography

With a career spanning 15 years at the forefront of Silicon Valley, Dr. Brian Foo brings a wealth of practical AI expertise in research and engineering. A graduate of UC Berkeley and UCLA, Dr. Foo has scaled AI technology at Uber and Google, yet his true passion lies in democratizing that knowledge. He is committed to ensuring the "AI revolution" doesn't leave diverse communities behind. By translating high-level research into accessible community growth and educational pathways, Brian aims to equip the next generation with the tools and opportunities they need to thrive in an AI-driven future.