deep_model.py

from for_prop import for_activation
def deep_model(X,parameters):
    """
    Takes X and parameters, respectively the data and the initialized parameters from init_param() function.
    Returns AV and caches, the activation value at the end of the architecture, and the cached values of every layer.
    """
    caches=[]
    A=X
    L=len(parameters)//2 #The floor division gives the number of layers in the network
    for l in range(1,L):
        A_prev=A
        A,cache=for_activation(A_prev,
                               parameters['W'+str(l)],
                               parameters['b'+str(l)],
                               activ='relu')
        caches.append(cache)
    AV,cache=for_activation(A,
                            parameters['W'+str(L)],
                            parameters['b'+str(L)],
                            activ='sigmoid')
    caches.append(cache)
    assert(AV.shape==(1,X.shape[1]))
    return AV,caches