Mastering TensorFlow 1.x

TFLearn-based MLP for MNIST classification

Now let's see how to implement the same MLP using TFLearn, another high-level library for TensorFlow:

1. Import the TFLearn library:

import tflearn

2. Define the hyper-parameters (we assume that the dataset has already been loaded into the X_train, Y_train, X_test, and Y_test variables):

num_layers = 2num_neurons = []for i in range(num_layers):num_neurons.append(256) learning_rate = 0.01n_epochs = 50batch_size = 100

3. Build the input layer, two hidden layers, and the output layer (the same architecture as examples in TensorFlow and Keras sections):

# Build deep neural networkinput_layer = tflearn.input_data(shape=[None, num_inputs])dense1 = tflearn.fully_connected(input_layer, num_neurons[0],  activation='relu') ...