"A Bayesian World"

[gusl]

This was one of the highlights from the NIPS workshops goodbye banquet:

[Error: unknown template video]

I was half-expecting the entire room of 200+ to join hands.

I saw them the night before and suggested a pun or two, but they rewrote most of it since then.

Comments

[spoonless.livejournal.com]

Fantastic!

Just reposted it to facebook :)

[gustavolacerda.livejournal.com]

as a physicist, did you ever study MCMC, Gibbs sampling, etc?

[spoonless.livejournal.com]

Hmm... this one is also good:

http://www.youtube.com/watch?v=aCIz7nje8w4

These not so much, but points for trying...

You Give Bayes a Bad Name:
http://www.youtube.com/watch?v=rfXd6tU6_uA

I'm In Love with a Bayesian:
http://www.youtube.com/watch?v=RJaC3t2O1x0

Markov Chain Monte Carlo:
http://www.youtube.com/watch?v=YhpxKAOd8_4

I had no idea this was a cottage industry!

[gustavolacerda.livejournal.com]

don't forget Brad Carlin's songbook.

[spoonless.livejournal.com]

We never talked about it in classes, but I used the Metropolis algorithm (and I think it would be an example of Gibbs sampling, or if not something very similar) to write some code to implement a (Markov Chain) Monte Carlo simulation which led to this paper...

http://prola.aps.org/abstract/PRB/v71/i17/e174438
http://arxiv.org/abs/cond-mat/0412150

It was the only work that I did in statistical mechanics; just took 1 summer, and led to a surprisingly fast paper (my first one). My subsequent papers were all in particle physics and very disconnected from it.

Actually, I never did understand what the application was of what we did to neural nets, even though we cross posted it to "disordered systems and neural networks" on arxiv.org. (it was not a disordered system, so presumably it has some application to neural nets; I'm just not sure what)

man, they keep on coming...

[spoonless.livejournal.com]

Like a Bayesian:
http://www.youtube.com/watch?v=LK_AS2y29cY&feature=related

We Didn't Start the Prior:
http://www.youtube.com/watch?v=ebiVF1OYXeo