Applied & Computational Mathematics seminar: A scenario of learning dynamics by reservoir computing

Speaker: Masato Hara, Kyoto University

Title:  A scenario of learning dynamics by reservoir computing

Abstract:  Reservoir computing is a machine learning method which can learn and predict deterministic dynamics. We are interested in the mathematical mechanism of reservoir computing, and we think that a reservoir becomes (semi) conjugate with a target system when the learning succeeds. In this talk I would like to introduce a scenario to explain how the conjugacy will be achieved. Here we use some classical techniques of structural stability theory, differential topology, etc. This is a joint work with Professor Hiroshi Kokubu (Kyoto University).

Time:  Monday, July 15 - 3:30pm - 4:30pm

Place:  Exploratory Hall, room 4106 and Zoom