"Rank these Agents" - Using Explanation to Reveal the Properties of a Set of Sequential Decision-making Agents
Complex systems, such as modern ML and AI techniques, have faced pervasive issues with transparency and assessment by humans. As a result, the widespread deployment of these systems have been shown to have unintended consequences, as it is very easy to overlook problems. This proposal outlines work leveraging how people come to understand and explain decisions, and builds upon it to support allowing ML non-experts to judge properties of a set of many sequential decision making agents via explanation. Improving the tools and techniques used to assess this type of agent will allow for improving assessment of a variety of ML and AI systems.
Co Advisor: Thomas Dietterich
Co Advisor: Margaret Burnett
Committee: Fuxin Li
Committee: Prasad Tadepalli
GCR: Henri Jansen
Thursday, January 30, 2020 at 10:00am to 11:55am
Kelley Engineering Center, 1114
110 SW Park Terrace, Corvallis, OR 97331
Calvin Hughes
No recent activity