python-kerasHow do I use Keras Sequential to create a model in Python?

Keras Sequential is a way of creating a linear stack of layers in a deep learning model. To use Keras Sequential, you must first create a Sequential object and then add layers to it. The layers can be added one by one or all at once.

Example code

from keras.models import Sequential
from keras.layers import Dense

# Create the Sequential model
model = Sequential()

# 1st Layer - Add a flatten layer
model.add(Flatten(input_shape=(32, 32, 3)))

# 2nd Layer - Add a fully connected layer
model.add(Dense(100, activation='relu'))

# 3rd Layer - Add a fully connected layer
model.add(Dense(60, activation='relu'))

# Output Layer - Add a fully connected layer
model.add(Dense(10, activation='softmax'))

# Compile the model
model.compile(optimizer='adam', loss='categorical_crossentropy', metrics=['accuracy'])

Code explanation

1. from keras.models import Sequential - imports the Sequential model from the Keras library.
2. model = Sequential() - creates the Sequential model object.
3. model.add(Flatten(input_shape=(32, 32, 3))) - adds a flatten layer to the model. The input shape is the size of the image (32x32 pixels with 3 color channels).
4. model.add(Dense(100, activation='relu')) - adds a fully connected layer with 100 nodes and ReLU activation.
5. model.add(Dense(60, activation='relu')) - adds another fully connected layer with 60 nodes and ReLU activation.
6. model.add(Dense(10, activation='softmax')) - adds the output layer with 10 nodes and Softmax activation.
7. model.compile(optimizer='adam', loss='categorical_crossentropy', metrics=['accuracy']) - compiles the model with the Adam optimizer, categorical cross-entropy loss function, and accuracy metrics.

Helpful links

• Keras Documentation - Sequential Model
• Keras Documentation - Layers
• Keras Documentation - Compilation

Edit this code on GitHub