Code-First Basic Deep Learning. Part of Code With Rahul.
(1) Start with MNIST-Logistic.ipynb
This introduces the ideas of softmax and gradient descent, and shows you how to do Logistic Regression using keras
.
(2) MNIST-MLP.ipynb
Introduces Neural Networs, specifically the Multi Layer Perceptron and shows how to fit it in Keras. Also introduces the ideas behind SGD.
(3) MNIST-MLP-Dropout.ipynb
The previous model is overfit. We need to fix this using a regularization technique called Dropout. We end with a general prescription on training neural networks