Help to verify Deep-Q-network, reinforcement learning algorithm source.

Revision en1, by rsampaths16, 2017-12-20 16:59:12

disclaimer : I do realise this request is not related to the conventional coding contests | problems asked here, but still hope to find people of similar interests, who would like to help.

I have been into Machine Learning and Deep Learning for a while now and it's been going smoothly. There is however a catch, that it's hard for me to verify the working of complex structures due to the limit's of my hardware compute (quad-core i5-processor, no Nvidia -> no Cuda). I've recently implemented Deep-Q-Network reinforcement learning algorithm, and it's taking eternity to converge (seriously it's hard to wait that long without results, or at-least a little sign that I was heading in the right direction). What I'm hoping from the community is that little sign, weather I'm heading in the right direction :) , or my implementation is wrong ;-; .

It would be cool if you could verify the source, and validate it. Even awesome if you could post results after training it.

Link to github repo : https://github.com/rsampaths16/dqn-game-bot clone github repo : git@github.com:rsampaths16/dqn-game-bot.git

I'm looking forward to any feedback you might have. You could give it here, even better as issues over there.

Thank You All;

Tags neural networks, deep learning, deep q rl, game agent

History

 
 
 
 
Revisions
 
 
  Rev. Lang. By When Δ Comment
en1 English rsampaths16 2017-12-20 16:59:12 1371 Initial revision (published)