tf_privacy.single_layer_softmax_classifier

Trains a single layer neural network classifier with softmax activation.

tf_privacy.single_layer_softmax_classifier(
    train_dataset: datasets.RegressionDataset,
    test_dataset: datasets.RegressionDataset,
    epochs: int,
    num_classes: int,
    optimizer: tf.keras.optimizers.Optimizer,
    loss: Union[tf.keras.losses.Loss, str] = 'categorical_crossentropy',
    batch_size: int = 32,
    kernel_regularizer: Optional[tf.keras.regularizers.Regularizer] = None
) -> Tuple[Any, List[float]]

Args
train_dataset	consists of num_train many labeled examples, where the labels are in {0,1,...,num_classes-1}.
test_dataset	consists of num_test many labeled examples, where the labels are in {0,1,...,num_classes-1}.
epochs	the number of epochs.
num_classes	the number of classes.
optimizer	a tf.keras optimizer.
loss	a tf.keras loss function.
batch_size	a positive integer.
kernel_regularizer	a regularization function.

Returns
List of test accuracies (one for each epoch) on test_dataset of model trained on train_dataset.