python-tensorflowHow do I use Python and TensorFlow to create a Convolutional Neural Network (CNN) example?

To create a Convolutional Neural Network (CNN) example using Python and TensorFlow, the following steps can be taken:

1. Import the necessary libraries:

import tensorflow as tf
from tensorflow.keras import datasets, layers, models

2. Load and prepare the MNIST dataset:

(train_images, train_labels), (test_images, test_labels) = datasets.mnist.load_data()

train_images = train_images.reshape((60000, 28, 28, 1))
test_images = test_images.reshape((10000, 28, 28, 1))

# Normalize pixel values to be between 0 and 1
train_images, test_images = train_images / 255.0, test_images / 255.0

3. Create the convolutional base:

model = models.Sequential()
model.add(layers.Conv2D(32, (3, 3), activation='relu', input_shape=(28, 28, 1)))
model.add(layers.MaxPooling2D((2, 2)))
model.add(layers.Conv2D(64, (3, 3), activation='relu'))
model.add(layers.MaxPooling2D((2, 2)))
model.add(layers.Conv2D(64, (3, 3), activation='relu'))

4. Add Dense layers on top:

model.add(layers.Flatten())
model.add(layers.Dense(64, activation='relu'))
model.add(layers.Dense(10, activation='softmax'))

5. Compile and train the model:

model.compile(optimizer='adam',
              loss='sparse_categorical_crossentropy',
              metrics=['accuracy'])

model.fit(train_images, train_labels, epochs=5)

6. Evaluate the model:

test_loss, test_acc = model.evaluate(test_images,  test_labels, verbose=2)

print(test_acc)

7. Output:

0.9909

Code Parts Explanation

• import tensorflow as tf: This imports the TensorFlow library into the Python environment.
• from tensorflow.keras import datasets, layers, models: This imports the datasets (e.g. MNIST), layers (e.g. Conv2D) and models (e.g. Sequential) modules from the Keras library.
• (train_images, train_labels), (test_images, test_labels) = datasets.mnist.load_data(): This loads the MNIST dataset into the environment.
• train_images = train_images.reshape((60000, 28, 28, 1)): This reshapes the training images into the shape required for the model (60000 images of 28x28 pixels with 1 channel).
• test_images = test_images.reshape((10000, 28, 28, 1)): This reshapes the test images into the shape required for the model (10000 images of 28x28 pixels with 1 channel).
• train_images, test_images = train_images / 255.0, test_images / 255.0: This normalizes the pixel values to be between 0 and 1.
• model = models.Sequential(): This creates a Sequential model.
• model.add(layers.Conv2D(32, (3, 3), activation='relu', input_shape=(28, 28, 1))): This adds a Conv2D layer with 32 filters of size 3x3, ReLU activation and input shape of 28x28x1.
• model.add(layers.MaxPooling2D((2, 2))): This adds a MaxPooling2D layer with pool size of 2x2.
• model.add(layers.Flatten()): This flattens the input.
• model.add(layers.Dense(64, activation='relu')): This adds a Dense layer with 64 nodes and ReLU activation.
• model.add(layers.Dense(10, activation='softmax')): This adds a Dense layer with 10 nodes and Softmax activation.
• model.compile(optimizer='adam', loss='sparse_categorical_crossentropy', metrics=['accuracy']): This compiles the model with Adam optimizer, Sparse Categorical Crossentropy loss and accuracy metric.
• model.fit(train_images, train_labels, epochs=5): This fits the model on the training data for 5 epochs.
• test_loss, test_acc = model.evaluate(test_images, test_labels, verbose=2): This evaluates the model on the test data.
• print(test_acc): This prints the accuracy of the model on the test data.

Relevant Links

• TensorFlow Tutorial - Convolutional Neural Network
• Keras Tutorial - Convolutional Neural Network

Edit this code on GitHub