Time & Location

2020. 11. 9. MON 7 PM

@KAIST, N1, Room #102

​

Moderator

Jea Kwon

​

Presenter

Jea Kwon
Ryungwoo Jang

​

Agenda

“What I cannot create, I do not understand”

- Richard Feynman

​

Backpropagation algorithm(backprop) enabled machines to learn deep neural net, which enabled humans to create state-of-the-art AIs. However, the biological plausibility of the backprop is questionable. In this session, we first visit the historical background why and how backprop has emerged. Next, we discuss on current consensus on biological plausibility of backprop and current research direction on biologically-plausible learning algorithm. Moreover, we introduce the current knowledge on biological learning rule and biologically-inspired learning algorithms.