Single Perceptron (Shallow Network)

single_perceptron.py

from sklearn.model_selection import train_test_split
from sklearn.metrics import confusion_matrix
from sklearn.preprocessing import LabelEncoder

import keras
from keras.models import Sequential
from keras.layers import Dropout, Flatten, Dense, BatchNormalization, Activation
from keras import metrics
from keras import backend as K

# Encode categorical data
labelencoder:LabelEncoder = LabelEncoder()
y = labelencoder.fit_transform(y)

# 80/20 split
X_train, X_test, y_train, y_test = train_test_split(X, y, test_size = 0.2, random_state = 0)

# GPU when available
K.tensorflow_backend._get_available_gpus()

# Initialize network
model:Sequential = Sequential()

# Single Perceptron/Shallow Network Architecture
model.add(Dense(use_bias=False, units=1, kernel_initializer="uniform", input_shape=(128, 30)))
# Regulate values going into each activation function
model.add(BatchNormalization())
# ELU activation 
# ref: https://ml-cheatsheet.readthedocs.io/en/latest/activation_functions.html#elu
model.add(Activation("elu"))

# Add dropout to prevent overfitting
model.add(Dropout(rate=0.1))

# Add output layer
model.add(Dense(use_bias=False, units=1, kernel_initializer="uniform"))
model.add(BatchNormalization())
model.add(Activation("sigmoid"))

# Summarize
model.summary()

# Define loss and optimizer 
model.compile(optimizer='adam', loss='categorical_crossentropy', metrics=['accuracy']) # binary_crossentropy

# Fit training set
model.fit(X_train, y_train, batch_size=100, epochs=150)

# Predicting on test set
y_pred = model.predict(X_test)
y_pred = (y_pred > 0.5)

# Confusion matrix
cm = confusion_matrix(y_test, y_pred)
print(cm)

# TODO: add references (confusion matrix), type hints