人工知能と機械学習における Java フレームワークの役割は何ですか?

import org.tensorflow.keras.models.Sequential;
import org.tensorflow.keras.layers.Dense;
import org.tensorflow.keras.layers.Conv2D;
import org.tensorflow.keras.layers.MaxPooling2D;
import org.tensorflow.keras.layers.Flatten;
import org.tensorflow.keras.optimizers.Adam;
import org.tensorflow.keras.datasets.mnist;
import org.tensorflow.keras.utils.ArrayUtils;

public class ImageClassifier {
    public static void main(String[] args) {
        // 加载 MNIST 数据集
        (x_train, y_train), (x_test, y_test) = mnist.load_data();
        
        // 预处理数据
        x_train = x_train.astype("float32") / 255;
        x_test = x_test.astype("float32") / 255;
        
        // 创建模型
        Sequential model = new Sequential();
        model.add(new Conv2D(32, (3, 3), activation="relu", input_shape=(28, 28, 1)));
        model.add(new MaxPooling2D((2, 2)));
        model.add(new Conv2D(64, (3, 3), activation="relu"));
        model.add(new MaxPooling2D((2, 2)));
        model.add(new Flatten());
        model.add(new Dense(128, activation="relu"));
        model.add(new Dense(10, activation="softmax"));
        
        // 编译模型
        model.compile(optimizer=new Adam(learning_rate=1e-3),
                loss="sparse_categorical_crossentropy",
                metrics=["accuracy"]);
        
        // 训练模型
        model.fit(x_train, y_train, epochs=5);
        
        // 评估模型
        System.out.println("准确率：" + model.evaluate(x_test, y_test)[1]);
    }
}