| Speaker: | Anton de la Fuente (Kavli IPMU) |
|---|---|
| Title: | Machine learning for people who already know all the math (Postdoc Colloquium) |
| Date (JST): | Fri, Dec 17, 2021, 11:00 - 11:30 |
| Place: | Zoom |
| Related File: | 2711.pdf |
| Abstract: | By learning from data, computers have learned to defeat world champions in chess, draw realistic pictures, and write human-sounding text. For physicists and mathematicians, the main idea in all these cases is actually conceptually quite simple. It is really an elaborate form of curve fitting. For example, some of the above tasks use curves containing over 100 billion tunable parameters. In this talk, I will explain how the above tasks can be viewed as curve-fitting problems, and I will describe the parameterization of the curves. |
| Contact: | https://ipmu.zoom.us/webinar/register/WN_VTTp-bOdQY6GHCMDQrFVIg |
| Remarks: | IPMU Postdoc Colloquium Series Registration necessary from here: https://ipmu.zoom.us/webinar/register/WN_VTTp-bOdQY6GHCMDQrFVIg |
