└── Handwritten Digit Classification


/Handwritten Digit Classification:
--------------------------------------------------------------------------------
 1 | from tensorflow.keras.datasets import mnist
 2 | from tensorflow.keras.models import Sequential
 3 | from tensorflow.keras.layers import Dense, Flatten, Conv2D, MaxPooling2D
 4 | from tensorflow.keras.callbacks import EarlyStopping
 5 | 
 6 | # Load the MNIST dataset
 7 | (X_train, y_train), (X_test, y_test) = mnist.load_data()
 8 | 
 9 | # Preprocess the data
10 | X_train = X_train.reshape(X_train.shape[0], 28, 28, 1).astype('float32') / 255
11 | X_test = X_test.reshape(X_test.shape[0], 28, 28, 1).astype('float32') / 255
12 | 
13 | # One-hot encode the labels
14 | y_train = pd.get_dummies(y_train).values
15 | y_test = pd.get_dummies(y_test).values
16 | 
17 | # Define the CNN model
18 | model = Sequential()
19 | model.add(Conv2D(32, (3, 3), activation='relu', kernel_initializer='he_uniform', input_shape=(28, 28, 1)))
20 | model.add(MaxPooling2D((2, 2)))
21 | model.add(Flatten())
22 | model.add(Dense(128, activation='relu', kernel_initializer='he_uniform'))
23 | model.add(Dense(10, activation='softmax'))
24 | 
25 | # Compile the model
26 | model.compile(loss='categorical_crossentropy', optimizer='adam', metrics=['accuracy'])
27 | 
28 | # Early stopping to prevent overfitting
29 | early_stopping = EarlyStopping(monitor='val_loss', patience=3)
30 | 
31 | # Train the model
32 | model.fit(X_train, y_train, epochs=10, validation_data=(X_test, y_test), callbacks=[early_stopping])
33 | 
34 | # Evaluate the model
35 | loss, accuracy = model.evaluate(X_test, y_test)
36 | print(f'Test accuracy: {accuracy:.2f}')
37 | 


--------------------------------------------------------------------------------