*********************************
There is now a CONTENT FREEZE for Mercury while we switch to a new platform. It began on Friday, March 10 at 6pm and will end on Wednesday, March 15 at noon. No new content can be created during this time, but all material in the system as of the beginning of the freeze will be migrated to the new platform, including users and groups. Functionally the new site is identical to the old one. webteam@gatech.edu
*********************************
Title: Interactive and Explainable Machine Learning For Humans
Date: April 18, 2022
Time: 12:00 – 2:00 PM Eastern Time
Location: Remote --
https://gatech.zoom.us/j/92102254383?pwd=ZFhkblZFbnVJcThXMDgzRXBnajFBUT09
Andrew Silva
Ph.D. Student in Computer Science
School of Interactive Computing
Georgia Institute of Technology
Committee:
Dr. Matthew Gombolay (Advisor), School of Interactive Computing, Georgia Institute of Technology
Dr. Sonia Chernova, School of Interactive Computing, Georgia Institute of Technology
Dr. Mark Riedl, School of Interactive Computing, Georgia Institute of Technology
Dr. Diyi Yang, School of Interactive Computing, Georgia Institute of Technology
Dr. Barry Theobald, Apple, Inc.
Abstract:
Interactivity and explainability within machine learning present an opportunity to improve human perceptions of AI, improve AI performance in complex domains, and improve human-AI teaming. While conventional machine learning research has focused on optimizing solely for an objective function (i.e., maximizing task completion rates, accuracy, or word-likelihood), virtual or embodied agents that are designed to interface with humans must be interactive and offer transparency to end-users. As machine learning is deployed in the real world, from smartphones to household robots, human users will benefit from personal- ization in their lives (e.g., predicting next words while considering an individual’s vernacular, or executing household tasks according to an individual user’s home layout or preference). Further, such agents must offer explanations and insights into their decision-making and behavior to improve human trust, understanding, and usability of such systems, and even to satisfy legal requirements. This thesis will present novel approaches to interaction and personalization in machine learning while simultaneously offering improved explainability for such systems. The result of this work will be the empowerment of human users to both (1) better-understand machine learning agents, and (2) tailor machine learning systems to their individual preferences.
