Reinforcement learning for Numberwang?

Attended some nice talks tonight at the London Machine Learning Meetup. First was Daniel Slater talking about his project “Alpha-Toe” (github here), which is reproducing the “Alpha-Go” system but for a variety of board games. It looks like an excellent way to learn about reinforcement learning. Then Piotr Mirowski of Google Deep Mind gave an easy-to-follow talk on recurrent neural networks, and their application to a variety of things, including game playing.

All of this game talk reminds me of the ultimate goal of designing an AI to play Numberwang. It won’t be likely to ever beat human players since the number of possible moves is 13+10^{|\mathbb{R}|}.


Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )


Connecting to %s