import tensorflow as tf
import numpy as np

def load_data():
    """
    Loads CIFAR-10 dataset and normalizes it.
    Returns:
        (x_train, y_train), (x_test, y_test)
    """
    (x_train, y_train), (x_test, y_test) = tf.keras.datasets.cifar10.load_data()
    
    # Normalize pixel values to be between 0 and 1
    x_train = x_train.astype('float32') / 255.0
    x_test = x_test.astype('float32') / 255.0
    
    return (x_train, y_train), (x_test, y_test)

def get_class_names():
    return ['airplane', 'automobile', 'bird', 'cat', 'deer',
            'dog', 'frog', 'horse', 'ship', 'truck']