Date: October 02, 2019 3:00pm-4:30pm
Location: Dunton Tower: Room 2203
Speaker: Andre Vellino
Title: Explanation and Artificial Neural Networks
Abstract:
As machine, learning systems become increasingly better than humans at making optimal choices, our ability and willingness to trust these choices will depend on their ability to also provide explanations for them. Explainable AI (XAI) aims to make machine decision making processes understandable to humans which will be a critical element in instruments on which we will increasingly depend – self-driving cars, medical diagnostic tools, fake news filtering etc.
Using AlphaGo’s victories against Go champion Lee Sedol in 2016 I will explore the question of what counts as an explanation for an ML-driven decision and what the feasible strategies are for synthesing such explanations.
Speaker’s profile:
André Vellino is an Associate Professor in Information Studies at the University of Ottawa and an Adjunct Research Professor at the Institute of Cognitive Science at Carleton University. He spent many years working in AI in the private sector and government.