Deep Learning and Research into Artifacts

Abstract

This document describes several topics on research into artifact from the perspective of deep learning. We first introduce the high-dimensional science proposed by Maruyama. Then, we explain the current research on model-based and model-free reinforcement learning, its integration, and world models. Then we bring the discussion by Yoshikawa about the academic domains and design. All the discussion is based on how the phenomenon is models either by a large number of parameters, or a small number of parameters which human can understand. Finally we discuss how the large-parameter models can be used in our society.