delta_rule#
- delta_rule(weights, feedback, input, alpha)[source]#
The change in the weights and the prediction errors of
cpm.models.learning.DeltaRule.- Parameters:
- weightsnumpy.ndarray
A 2D array, one row per outcome.
- feedbacknumpy.ndarray
The teaching signal, one per outcome.
- inputnumpy.ndarray
The stimulus representation, one value per column of
weights.- alpha
float The learning rate.
- Returns:
- changenumpy.ndarray
The change in
weights, with the same shape.- errornumpy.ndarray
The summed prediction error of each outcome.