[MINI] Dropout

Deep learning can be prone to overfit a given problem. This is especially frustrating given how much time and computational resources are often required to converge. One technique for fighting overfitting is to use dropout.