OmegaGo

Thursday February 16, 2017

This is a little fun presentation for DC Hack and Tell on Thursday, February 16, 2017. (slides)

Hi! I'm Aaron. This is my blog and my twitter handle. You can get from one to the other. This presentation and a corresponding write-up (you're reading it) are on my blog (which you're on).

Here is a very serious logo.

(It's based on ¯\_(ツ)_/¯, if that wasn't clear.)

Here's what Go sometimes looked like in Pi.

Go really looks more like this. It's a game, see?

xkcd had this cartoon about game AIs and Go was on it.

Go was supposed to be hard for computers to be good at.

But... AlphaGo.

TWO THINGS

First technically interesting things, and then a personally interesting thing.

Neural nets! Policy networks! Value networks!

Interesting, but/and I do that kind of thing for work.

(Figure from the Nature article.)

David Silver's reinforcement learning course.

Okay tree search is the part I'm interested in.

(Figure from the Nature article.)

me: 0

computer: ∞

The other thing is that I suck at Go, so I don't need/want a Go computer to be so spanking good.

Here is a very serious logo.

code!

Talk about code here. It's on github.

Here is a very serious logo.

demo!

Demo that code from up there!

Thanks!

Thank you!

This is just me again.