Deep-Learning-algorithms

Implementation of all basic algorithms needed in Deep Learning

1. Logistic_regression_using_NN.ipynb

   • It is a simple logistic regression algorithm developed using NN (Neural Networks) with zero hidden layers
   • In this notebook, binary classification is done on the dataset of cats(cat or not cat)

2. NN_with_1_hidden_layer.py

   • It's a python script file which contains all functions required to develop NN (Neural Networks) with one hidden layer
   • function designed are:
     • sigmoid(z)
     • initialize(n_x,n_h1,n_y)
     • forword_propagation(X,parameters)
     • evaluate_cost(A2,Y, parameters, lambd)
     • backword_propagation(X,Y,cache,parameters,lambd)
     • update_parameters(parameters,grads,learning_rate)
     • predict(parameters, X)
     • model(X_train, Y_train, X_test, Y_test,n_h1, num_iterations, learning_rate,lambd)
     • plot_cost(costs)

3. deep_NN_with_L_layers.py

   • It's a python script file which contains all functions required to develop NN (Neural networks) with 'n' hidden layer
   • It's the generalized algorithm for CNN
   • Functions designed are:
     • sigmoid(Z)
     • relu(Z)
     • sigmoid_backward(dA, cache)
     • relu_backward(dA, cache)
     • initialize_parameters(layer_dims)
     • linear_forward(A, W, b)
     • linear_activation_forward(A_prev, W, b, activation)
     • L_model_forward(X, parameters)
     • compute_cost(AL, Y)
     • linear_backward(dZ, cache)
     • linear_activation
     • linear_activation_backward(dA, cache, activation)
     • L_model_backward(AL, Y, caches)
     • update_parameters(parameters, grads, learning_rate)
     • predict(X, parameters)
     • L_layer_model(X, Y, layers_dims, learning_rate, num_iterations, print_cost)
     • plot_cost(costs)