ICPS
2021 APS Virtual Convention · 2021
Opening the Opaque-Box: Strength, Limitations, and Recommendations in the Application of Scientific Regret Minimization to Behavioral Modeling
- Alexandre Filipowicz
Toyota Research Institute - Totte Harinen
Toyota Research Institute - Rumen Iliev
Toyota Research Institute - Yanxia Zhang
Toyota Research Institute - Yan-Ying Chen
Toyota Research Institute - Yue Weng
Toyota Research Institute - Abishek Komma
Toyota Research Institute - Kent Lyons
Toyota Research Institute - Charlene Wu
Toyota Research Institute
Abstract
The ‘Scientific Regret Minimization’ method proposes steps to use machine learning in the development of precise, interpretable models of behavior. Using simulations, we highlight strengths and limitations with this method and offer solutions to leverage the predictive power of machine learning in the development and improvement of behavioral models.
Decision Making