Profile cover photo
Profile photo
Tambet Matiisen
free electron
free electron
About
Tambet's posts

Post has attachment

Post has attachment
I wrote an accessible introduction to deep reinforcement learning (i.e. how DeepMind's code plays Atari games):

http://www.nervanasys.com/demystifying-deep-reinforcement-learning/

I tried to keep the text simple, but still cover all the key points. Did it work for you?

Post has attachment

Post has attachment

Post has attachment
I'm releasing my own implementation of deep Q-learning, used in DeepMind's seminal paper to play Atari games. My code was designed to be simple, fast and easy to extend, so I believe it is particularly suitable for anyone starting in this area. It is not yet matching DeepMind's results 100%, but is close enough. I would appreciate any help with testing to bring it on-par with their results.

Give it a try:
https://github.com/tambetm/simple_dqn

Post has attachment
Wait while more posts are being loaded