Feedforward Neural Network For Binary Classification

Preliminaries

# Load libraries
import numpy as np
from keras.datasets import imdb
from keras.preprocessing.text import Tokenizer
from keras import models
from keras import layers

# Set random seed
np.random.seed(0)
Using TensorFlow backend.


# Set the number of features we want
number_of_features = 1000

# Load data and target vector from movie review data
(train_data, train_target), (test_data, test_target) = imdb.load_data(num_words=number_of_features)

# Convert movie review data to one-hot encoded feature matrix
tokenizer = Tokenizer(num_words=number_of_features)
train_features = tokenizer.sequences_to_matrix(train_data, mode='binary')
test_features = tokenizer.sequences_to_matrix(test_data, mode='binary')

Construct Neural Network Architecture

Because this is a binary classification problem, one common choice is to use the sigmoid activation function in a one-unit output layer.

# Start neural network
network = models.Sequential()

# Add fully connected layer with a ReLU activation function

# Add fully connected layer with a ReLU activation function

# Add fully connected layer with a sigmoid activation function
network.add(layers.Dense(units=1, activation='sigmoid'))

Compile Feedforward Neural Network

# Compile neural network
network.compile(loss='binary_crossentropy', # Cross-entropy
optimizer='rmsprop', # Root Mean Square Propagation
metrics=['accuracy']) # Accuracy performance metric

Train Feedforward Neural Network

In Keras, we train our neural network using the fit method. There are six significant parameters to define. The first two parameters are the features and target vector of the training data.

The epochs parameter defines how many epochs to use when training the data. verbose determines how much information is outputted during the training process, with 0 being no out, 1 outputting a progress bar, and 2 one log line per epoch. batch_size sets the number of observations to propagate through the network before updating the parameters.

Finally, we held out a test set of data to use to evaluate the model. These test features and test target vector can be arguments of the validation_data, which will use them for evaluation. Alternatively, we could have used validation_split to define what fraction of the training data we want to hold out for evaluation.

In scikit-learn fit method returned a trained model, however in Keras the fit method returns a History object containing the loss values and performance metrics at each epoch.

# Train neural network
history = network.fit(train_features, # Features
train_target, # Target vector
epochs=3, # Number of epochs
verbose=1, # Print description after each epoch
batch_size=100, # Number of observations per batch
validation_data=(test_features, test_target)) # Data for evaluation
Train on 25000 samples, validate on 25000 samples
Epoch 1/3
25000/25000 [==============================] - 2s - loss: 0.4215 - acc: 0.8102 - val_loss: 0.3385 - val_acc: 0.8558
Epoch 2/3
25000/25000 [==============================] - 1s - loss: 0.3241 - acc: 0.8646 - val_loss: 0.3261 - val_acc: 0.8626
Epoch 3/3
25000/25000 [==============================] - 2s - loss: 0.3120 - acc: 0.8700 - val_loss: 0.3268 - val_acc: 0.8593