Merge Chapter13

MLP Tuning with Keras Tuner

Author: jElhamm

Chapter13/chapter_13.py

@@ -0,0 +1,127 @@
+import numpy as np
+import tensorflow as tf
+import keras_tuner as kt
+from tensorflow import keras
+from tensorflow.keras import layers
+
+
+class SimpleMLP(kt.HyperModel):
+    def __init__(self, num_classes):
+        self.num_classes = num_classes
+
+    def build(self, hp):
+        units = hp.Int(name="units", min_value=16, max_value=64, step=16)
+        optimizer = hp.Choice(name="optimizer", values=["rmsprop", "adam"])
+
+        model = keras.Sequential([
+            layers.Dense(units, activation="relu"),
+            layers.Dense(self.num_classes, activation="softmax")
+        ])
+
+        model.compile(
+            optimizer=optimizer,
+            loss="sparse_categorical_crossentropy",
+            metrics=["accuracy"]
+        )
+        return model
+
+class HyperparameterTuner:
+    def __init__(self, hypermodel, max_trials=100, executions_per_trial=2, directory="mnist_kt_test"):
+        self.tuner = kt.BayesianOptimization(
+            hypermodel,
+            objective="val_accuracy",
+            max_trials=max_trials,
+            executions_per_trial=executions_per_trial,
+            directory=directory,
+            overwrite=True
+        )
+
+    def search(self, x_train, y_train, x_val, y_val):
+        callbacks = [keras.callbacks.EarlyStopping(monitor="val_loss", patience=5)]
+
+        self.tuner.search(
+            x_train, y_train,
+            batch_size=128,
+            epochs=100,
+            validation_data=(x_val, y_val),
+            callbacks=callbacks,
+            verbose=2
+        )
+
+    def get_best_hyperparameters(self, top_n=4):
+        return self.tuner.get_best_hyperparameters(top_n)
+
+    def get_best_models(self, top_n=4):
+        return self.tuner.get_best_models(top_n)
+
+class ModelTrainer:
+    def __init__(self, x_train_full, y_train_full, x_train, y_train, x_val, y_val):
+        self.x_train_full = x_train_full
+        self.y_train_full = y_train_full
+        self.x_train = x_train
+        self.y_train = y_train
+        self.x_val = x_val
+        self.y_val = y_val
+
+    def get_best_epoch(self, model_builder, hp):
+        model = model_builder.build(hp)
+        callbacks = [keras.callbacks.EarlyStopping(monitor="val_loss", mode="min", patience=10)]
+
+        history = model.fit(
+            self.x_train, self.y_train,
+            validation_data=(self.x_val, self.y_val),
+            epochs=100,
+            batch_size=128,
+            callbacks=callbacks
+        )
+
+        val_loss_per_epoch = history.history["val_loss"]
+        best_epoch = val_loss_per_epoch.index(min(val_loss_per_epoch)) + 1
+        print(f"Best epoch: {best_epoch}")
+        return best_epoch
+
+    def get_best_trained_model(self, model_builder, hp):
+        """آموزش مدل با بهترین تعداد epoch."""
+        best_epoch = self.get_best_epoch(model_builder, hp)
+        model = model_builder.build(hp)
+
+        model.fit(
+            self.x_train_full, self.y_train_full,
+            batch_size=128,
+            epochs=int(best_epoch * 1.2)
+        )
+        return model
+
+def prepare_data():
+    (x_train, y_train), (x_test, y_test) = keras.datasets.mnist.load_data()
+    x_train = x_train.reshape((-1, 28 * 28)).astype("float32") / 255
+    x_test = x_test.reshape((-1, 28 * 28)).astype("float32") / 255
+
+    x_train_full = x_train[:]
+    y_train_full = y_train[:]
+
+    num_val_samples = 10000
+    x_train, x_val = x_train[:-num_val_samples], x_train[-num_val_samples:]
+    y_train, y_val = y_train[:-num_val_samples], y_train[-num_val_samples:]
+
+    return x_train_full, y_train_full, x_train, y_train, x_val, y_val, x_test, y_test
+
+
+
+keras.mixed_precision.set_global_policy("mixed_float16")
+x_train_full, y_train_full, x_train, y_train, x_val, y_val, x_test, y_test = prepare_data()
+
+hypermodel = SimpleMLP(num_classes=10)
+tuner = HyperparameterTuner(hypermodel)
+tuner.search(x_train, y_train, x_val, y_val)
+
+best_hps = tuner.get_best_hyperparameters(top_n=4)
+trainer = ModelTrainer(x_train_full, y_train_full, x_train, y_train, x_val, y_val)
+
+best_models = []
+for hp in best_hps:
+    model = trainer.get_best_trained_model(hypermodel, hp)
+    model.evaluate(x_test, y_test)
+    best_models.append(model)
+
+best_models = tuner.get_best_models(top_n=4)