Machine/Deep Learning Theory + Physics Seminar

(This seminar has been discontinued since Mar 2023.)

Held in cooperation with KAIST AI (OSI Lab, OptiML Lab), KAIST Physics (CSSPL), and KIAS, this is a special topics paper seminar on machine/deep learning theory and physics. At each seminar, two people present paper(s) of relevant topics. The seminar is held hybrid (in-person for people at the Seoul campus of KAIST, online for others). This is advised by Prof. Se-Young Yun of OSI Lab, Prof. Chulhee Yun of OptiML Lab, and Prof. Hawoong Jeong of CSSPL.

Refer to the website for more details: https://sites.google.com/view/mdlt-p

Gradient Descent on Infinitely Wide Neural Networks

Event Weekly DL Theory & Stat Phy Seminar Short summary In this seminar, I will talk about the paper “Gradient Descent on Infinitely Wide Neural Networks: Global Convergence and …

Oct 17, 2022 • 1 min read

Landscape and training regimes in deep learning

Event Weekly DL Theory & Stat Phy Seminar Short summary In this seminar, I will talk about a line of works that tries to explain the phase transition behavior of the loss landscape …

Sep 6, 2022 • 1 min read

Entropic variants of SGD

Event Weekly DL Theory & Stat Phy Seminar Short summary In this seminar, I will talk about a line of works that propose a new definition of “flatness” for loss landscapes in terms …

Jul 11, 2022 • 1 min read

Continuous Heavy-Tailed Theory of SGD

Event Weekly DL Theory & Stat Phy Seminar Short summary In this seminar, I will talk about a recent line of works that propose to analyze SGD under heavy-tail noise assumptions …

Jul 4, 2022 • 1 min read

No results found

Machine/Deep Learning Theory + Physics Seminar

Gradient Descent on Infinitely Wide Neural Networks

Landscape and training regimes in deep learning

Entropic variants of SGD

Continuous Heavy-Tailed Theory of SGD