tag:blogger.com,1999:blog-4446292666398344382.post3130113581492149033..comments2022-11-20T13:12:04.203-08:00Comments on Machined Learnings: Interactive PACPaul Mineirohttp://www.blogger.com/profile/05439062526157173163noreply@blogger.comBlogger4125tag:blogger.com,1999:blog-4446292666398344382.post-77527410219545041872012-05-10T13:27:21.821-07:002012-05-10T13:27:21.821-07:00definitely!definitely!Lev Reyzinhttps://www.blogger.com/profile/09629175455869565423noreply@blogger.comtag:blogger.com,1999:blog-4446292666398344382.post-29320091060566745962012-05-09T22:11:25.511-07:002012-05-09T22:11:25.511-07:00Well I clearly need to chew on your paper for a wh...Well I clearly need to chew on your paper for a while, but the overall point is immediately apparent: you can use martingales in an adversarial setting to prove results about your own randomizations.<br /><br />Viva la martingale!Paul Mineirohttps://www.blogger.com/profile/05439062526157173163noreply@blogger.comtag:blogger.com,1999:blog-4446292666398344382.post-73412130859721472712012-05-09T21:15:37.526-07:002012-05-09T21:15:37.526-07:00Thanks for pointing that out, I'll put that on...Thanks for pointing that out, I'll put that on my reading list.<br /><br />Re: game theory, I may have been too hasty in declaring the future to be all game theory :) proposition 2 of http://arxiv.org/abs/1104.5070 indicates for OCO-style problems there is always a minimax strategy for the adversary which is oblivious (but nonstationary). So maybe statistical reasoning will be sufficient to master these problems.Paul Mineirohttps://www.blogger.com/profile/05439062526157173163noreply@blogger.comtag:blogger.com,1999:blog-4446292666398344382.post-26955900129203804982012-05-09T19:09:26.498-07:002012-05-09T19:09:26.498-07:00Ran into your blog. You might be interested in an...Ran into your blog. You might be interested in another application of a Freedman-style inequality (http://arxiv.org/abs/1002.4058). Here, the data can be adversarial (e.g. not necessarily iid), but the algorithm can act randomly and still do well. This doesn't get into game theory, however, because of how regret is defined -- this may be a limitation of the model.Lev Reyzinhttps://www.blogger.com/profile/09629175455869565423noreply@blogger.com