Autoencoder

Autoencoder class

docclean.autoencoder.Autoencoder(
    img_rows=256, img_cols=256, nchans=3, mixed_precision_training=False, early_stopping=25
)

Autoencoder class.

Args:

img_rows (int): Number of pixel rows

img_cols (int): Number of pixel columns

nchans (int): Number of color channels

mixed_precision (bool): Train with fp16 mixed precision

early_stopping (int): Patience for early stopping

---

build_model method

Autoencoder.build_model()

Build the autoencoder model

Returns:

tf.keras.Model: The autoencoder model

---

train_model method

Autoencoder.train_model(train_dataset, epochs=100, validation_data=None)

Train the autoencoder

Args:

train_dataset (tf.data.Dataset) : Training dataset

epochs (int): Number of epochs to train for

validation_data (tf.data.Dataset) : Validation dataset

Returns:

dict: Training history

---