Parallel Learning between Science for AI and AI for Science: A Brief Overview and Perspective*

Abstract

This paper gives a brief overview of how AI and Science promote each other. First, the main algorithms inspired by mathematics, physics, and biology during the last 80~ years are summarized. Second, on the counter-direction, AI methods that help to solve hard problems in recent years are introduced. Perspectives and challenges are also discussed. Third, the Parallel Learning is introduced for the purpose of a general framework to assist the development of both Science for AI and AI for Science.