Using RL to solve overfitting in neural networks - View it on GitHub