ghazaltariri / Building-Deep-Neural-Network

Building NN from scratch

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

Building-Deep-Neural-Network

Building NN from scratch: These are the methods for building NN:

  1. initialize_parameters_deep(layer_dims) which returns parameters. We should make sure that dimensions match between each layes.
  2. L_model_forward(X, parameters) which return AL, caches
  3. compute_cost(AL, Y) which return cost
  4. L_model_backward(AL, Y, caches) which return grads
  5. update_parameters(parameters, grads, learning_rate) which return parameters I will use two activation functions:

Sigmoid:

alt text

This function returns two items: the activation value "a" and a "cache" that contains "Z"

ReLU:

A = RELU(Z) = max(0, Z)

The model's structure is : [Linear-> Relu] X (L-1) -> Linear -> Sigmoid. I.e., it has L-1 layers using ReLU activation function followed by an output layer with a sigmoid activation function.

We can use random initialization for the weight metrics and zero initialization for biases. After initialization, we will do the forward propagation module by implementing some basic functions: -Linear -Linear -> Activation where Activation will be either ReLU or Sigmoid -[Linear-> Relu] X (L-1) -> Linear -> Sigmoid

About

Building NN from scratch

License:MIT License


Languages

Language:Python 100.0%