Top.Mail.Ru
? ?
Notes from a Medium-Sized Island [entries|archive|friends|userinfo]
Jason

[ website | My Website ]
[ userinfo | livejournal userinfo ]
[ archive | journal archive ]

[Mar. 2nd, 2005|07:54 pm]
Jason
[Tags|, , ]

Kleinberg class today was extra good today. It was a great explanation of spectral clustering --- what I sort of understood from Bar-Joseph and Moore's machine learning class I know understand much better. Much like the whole topic of generating functions, it's attractive partly because it's so mysteriously effective. It's the kind of mystery that suggests there may be some very nice theorems left to be proven.
LinkReply

Comments:
From: rweba
2005-03-03 01:10 am (UTC)
Generating functions are not understood?

I know I don't understand them, but I would expect that the math guys would have been thinking about them for a long time and solved all the problems.
(Reply) (Thread)
[User Picture]From: jcreed
2005-03-03 01:17 am (UTC)
Well, I should have included "and their connection to type theory", but I'm not sure how well understood they are even as a bridge between combinatorics and calculus. You're right that they're well understood enough that they've received heavy use for decades, but to my understanding there's still a little bit of "okay, well why is this actually working?" mystery left.
(Reply) (Parent) (Thread)
[User Picture]From: jcreed
2005-03-03 08:27 pm (UTC)
Awesome! Thanks for the link.
(Reply) (Parent) (Thread)