rserizel / TGNMF

Supervised nonnegative matrix factorisation
2 stars 2 forks source link

binary classification TNMF #1

Open nayfous opened 5 years ago

nayfous commented 5 years ago

Is there a way of extending TNMF to work for binary classification? The code now expects the classifier to return a coefficient vector for every label and fails at line 333. The logistic regression function returns only 1 vector when there are 2 classes.

rserizel commented 5 years ago

Please find below a version that seems to be working (note that it will work only for binary LR). Binary_TNMF.zip

Please let me know if this is working in your case.

Best, Romain

nayfous commented 5 years ago

thank you, it works for me