Skip to content

Instantly share code, notes, and snippets.

@zerolocker

zerolocker/《预测、学习和博弈》学习笔记

Created September 23, 2014 03:18
Show Gist options
  • Star 0 You must be signed in to star a gist
  • Fork 0 You must be signed in to fork a gist
  • Save zerolocker/adcb56615019b962ccf1 to your computer and use it in GitHub Desktop.
Save zerolocker/adcb56615019b962ccf1 to your computer and use it in GitHub Desktop.
Download ZIP
《预测、学习和博弈》读书笔记
Raw
《预测、学习和博弈》学习笔记
Prediction,Learning,and Games
========================
+ Idea from P14: Would it be possible to select some weight vectors which have smaller loss function values when training Algorithm 2, and average those weight vectors $w$ *in some manner to be designed* so as to obtain a more accurate/better weight vector $w$?
> Written with [StackEdit](https://stackedit.io/).
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment