Incorrect Perceptron default values for sklearn models
In Fed-BioMed, Perceptron
model is currently created out of the SGDClassifier
model by forcing the nature of the loss function (loss=perceptron
). While mathematically correct, this way of doing things leads to some incorrect default values in Perceptron
model, that are different to the default values of SGDClassifier
(see Perceptron and SGDClassifier API documentation).
This issue has been open to address this issue: making sure initial default values are the same as the Perceptron
model