Deep Q-learning

Now that we understand some of the foundations of neural networks, it will be really helpful to look at a very basic example in Python that demonstrates their use. Go through the following steps to build a neural network that trains an agent with deep Q-learning. Windows users, make sure that you open an Anaconda prompt and switch to the ml-agents environment with activate ml-agents, as we did earlier.

1. Open a command prompt or shell window to an empty folder and enter the following:

git clone https://github.com/matthiasplappert/keras-rl.git
      cd keras-rl
      python setup.py install

2. This will install Keras RL, the reinforcement learning package for Keras. Keras is a popular library for building neural networks and other ML tasks. It can be backed by TensorFlow or Theano. Since we already have TensorFlow installed, we are good.
3. Enter the following:

      pip install h5py
      pip install gym

4. The h5py code phrase is a Pythonic interface to the HDF5 format, and allows you to...