pyconstruct.learners.BlockCoordinateFrankWolfe.fit

BlockCoordinateFrankWolfe.fit(X, Y, Y_pred=None)

Updates the current model with a mini-batch (X, Y).

Parameters:

• X (numpy.ndarray) – Input examples. The first dimension must be the batch size.
• Y (numpy.ndarray) – Output objects. The first dimension must be the batch size. This must coincide with batch size for X.
• Y_pred (numpy.ndarray) – Predictions of the algorithm. The first dimension must be the batch size. This must coincide with batch size for X. If None, either not needed or done internally.

Returns:

Return type:

self