def predictions_to_classes(self,
predictions: tf.Tensor,
best=True) -> pd.DataFrame:
if best:
return pd.DataFrame({
"class_id": predictions.argmax(-1),
"score": predictions.max(-1)
}).assign(
label=lambda df: df.class_id.map(self.id_to_label.__getitem__))
return pd.DataFrame(predictions, columns=self.id_to_label)
def fit(mps: classifier.MatrixProductState,
optimizer,
x: tf.Tensor,
y: tf.Tensor,
x_val: Optional[tf.Tensor] = None,
y_val: Optional[tf.Tensor] = None,
n_epochs: int = 20,
batch_size: int = 10,
n_message: int = 1):
"""Supervised training of an MPS classifier on a dataset.
Args:
mps: MatrixProductState classifier object.
optimizer: TensorFlow optimizer object to use in training.
A working option is AdamOptimizer with learning_rate=1e-4.
x: Training data (encoded images) of shape (n_data, n_sites, d_phys)
y: Training labels in one-hot format of shape (n_data, n_labels)
x_val: Validation data to calculate loss and accuracy during training.
y_val: Validation labels to calculate loss and accuracy during training.
n_epochs: Total number of epochs to train.
batch_size: Batch size for training.
n_message: Every how many epoch to print messages (loss, accuracy, times).
"""
data = tf.cast(x, dtype=mps.dtype)
labels = tf.cast(y, dtype=mps.dtype)
n_batch = len(x) // batch_size
if x_val is not None:
data_val = tf.cast(x_val, dtype=mps.dtype)
labels_val = tf.cast(y_val, dtype=mps.dtype)
n_batch_val = len(x_val) // batch_size
history = {
"loss": [],
"acc": [],
"total_time": [],
"val_loss": [],
"val_acc": []
}
start_time = time.time()
for epoch in range(n_epochs):
generator = ((data[i * batch_size:(i + 1) * batch_size],
labels[i * batch_size:(i + 1) * batch_size])
for i in range(n_batch))
loss, logits = run_epoch(mps, generator, n_batch, optimizer)
history["loss"].append(loss / len(x))
history["acc"].append(
(logits.numpy().argmax(axis=1) == y.argmax(axis=1)).mean())
history["total_time"].append(time.time() - start_time)
if x_val is not None:
val_generator = ((data_val[i * batch_size:(i + 1) * batch_size],
labels_val[i * batch_size:(i + 1) * batch_size])
for i in range(n_batch_val))
val_loss, val_logits = run_epoch(mps, val_generator, n_batch_val)
history["val_loss"].append(val_loss / len(x_val))
history["val_acc"].append((val_logits.numpy().argmax(
axis=1) == y_val.argmax(axis=1)).mean())
if epoch % n_message == 0:
print("\nEpoch: {}".format(epoch))
print("Time: {}".format(history["total_time"][-1]))
print("Loss: {}".format(history["loss"][-1]))
print("Accuracy: {}".format(history["acc"][-1]))
if x_val is not None:
print("Validation Accuracy: {}".format(history["val_acc"][-1]))
return mps, history