This tutorial will aim to give the audience a brief methodological survey of the two-way interaction between the study of human behavior and data science. In the first part of the tutorial, I will present an example of how cognitive science research can potentially improve machine learning methods. Specifically, I will describe how human similarity judgments can be used to improve visual representations learned by conventional deep neural networks. In the second, I will present an example of how machine learning tools can be used to achieve a richer scientific understanding of human behavior. Specifically, I will present a hierarchical model of visual information search that yokes multiple parallel stochastic accumulator processes to a TD-learning meta-controller and show how it explains humans' decisions to terminate information search better than existing models in cognitive science.
Assistant Professor, Computer Science & Engineering, Indian Institute of Technology, Kanpur