Cat vs Non-cat Classifier - Defining some utility functions

As a workflow, we shall first define the helper functions to be used in our holistic algorithm.

We are going to define the following functions.

Sigmoid Function - calculates the sigmoid activation function
Initialize Weights - initializes the weights and bias values to zeros, upon which we update those values through optimization
Forward propagation - returns activations and cost
Backward propagation - returns the gradients(derivatives) of weights and bias
Propagate - implements the forward and backward propagation by calling the corresponding functions
Predict - predicts the labels
Get Accuracies - calculates the accuracies of predictions of the model
Optimize - gets the optimal weights and biases
Model - the main function to train the algorithm and get the final model for classification