Skip to main content

Binary Classification Second Attempt

Summary​

  • Adjusting the linear model for non-linear problem

Content​

Adjusting the linear model for non-linear problem​

  • In a binary classification problem, we can use ReLU for hidden layers, and sigmoid for output layers
model = tf.keras.Sequential([
    tf.keras.layers.Input(2),
    tf.keras.layers.Dense(10, activation = 'relu'),
    tf.keras.layers.Dense(10, activation = 'relu'),
    tf.keras.layers.Dense(1, activation = 'sigmoid'),
])

model.compile(
    loss = tf.keras.losses.BinaryCrossentropy(),
    optimizer = tf.keras.optimizers.Adam(),
    metrics = ['accuracy']
)

model.fit(X_train, y_train, epochs = 200, verbose=0)

![[Pasted image 20231227131315.png]]