You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
The OneVsAll softmax probability calibration method in shogun currently uses parameters learnt on a sigmoid. A method is needed that can be used to learn parameters for a softmax function instead.
The OneVsAll softmax probability calibration method in shogun currently uses parameters learnt on a sigmoid. A method is needed that can be used to learn parameters for a softmax function instead.
The current
fit_sigmoid
callshogun/src/shogun/machine/MulticlassMachine.cpp
Line 132 in d1763b8
should be replaced by some method using which we can learn params for softmax.
See description of equation 7 in paper https://hal.inria.fr/inria-00103955/document.
The text was updated successfully, but these errors were encountered: