Neural-Network-from-scratch

Summary of Training a Neural Network

1. Randomly initialize the weights.
2. Implement forward propagation to get h_θ(x_i) for any x_i.
3. Implement the cost function.
4. Implement backpropagation to compute partial derivatives of weights and biases.
5. Use gradient descent to minimize the cost function with weights in θ.

My Words

I first learned neural network algorithm from a free online ML course series by Andrew Ng. Mr. Ng did a very great job on explaining every detail, and I had a general understanding of neural network. Nonetheless, I still tried to build neural network from scratch, and this exercise indeed helps me gain a much deeper comprehension of how different algorithms, such as backpropagation and gradient descent, come together to build a neural network.