Deep Learning: Architecture & Examples
Prepare Data - load, analysis, feature engineering, normalization…
Identify the problem - regression or classification
Define a neural network architecture
Create a neural network according to an architecture
Initialize the model - Sequential
Create the input layer
Using Flatten
During the creation of the first hidden layer
Create the hidden layers - Dense
Initialize the parameters
kernel_initializer - random_uniform, he_normal…
bias_initializer - zeors, Constant(0.05)
Define the activation functions: linear, sigmoid, relu…
Create the output layer
Initialize the parameters
Define the activation function
Compile the model - compile(Optimizer, loss, metrics…)
Train the model - fit(X, y, epochs, batch_size…)
Make predictions - predict(X)
Evaluate the model