Python Neural Network Reinforcement Learning [closed]

Question

I want to make a Neural Network that is trained using reinforcement learning in python.

X -> [ANN] -> yEstimate -> score! -> (repeat until weights are optimised)

I'm using Scikit-learn at the moment but there doesn't seem to be all the neural networks stuff tries to fit yEstimate to yTarget.

Are there secrets to scikit-learn or are there other libraries that I don't know about for accomplishing this?

Thank you!

Answer 1

Google TensorFlow is one big library with a Python interface for the creation of neural networks. See https://github.com/nivwusquorum/tensorflow-deepq for more information and examples about reinforcement learning.

Answer 2

Here you have an example http://pybrain.org/

Here another https://pypi.python.org/pypi/Reinforcement-Learning-Toolkit/1.0

There are many research groups in robotics using reinforcement learning. If you don't like the ones that I gave you, you can try to look in these groups