def mean_squared_error_regressor(tensor_in, labels, weights, biases, name=None):
"""Returns prediction and loss for mean squared error regression."""
with ops.op_scope([tensor_in, labels], name, "mean_squared_error_regressor"):
predictions = nn.xw_plus_b(tensor_in, weights, biases)
if len(labels.get_shape()) == 1 and len(predictions.get_shape()) == 2:
predictions = array_ops_.squeeze(predictions, squeeze_dims=[1])
return predictions, loss_ops.sum_of_squares(predictions, labels)
def mean_squared_error_regressor(tensor_in,
labels,
weights,
biases,
name=None):
"""Returns prediction and loss for mean squared error regression."""
with ops.op_scope([tensor_in, labels], name,
"mean_squared_error_regressor"):
predictions = nn.xw_plus_b(tensor_in, weights, biases)
if len(labels.get_shape()) == 1 and len(predictions.get_shape()) == 2:
predictions = array_ops_.squeeze(predictions, squeeze_dims=[1])
return predictions, loss_ops.sum_of_squares(predictions, labels)