computer vision - Machine Learning - SVM - How to calculate bias when calculate vector W? -
i writing code of svm primal uses sgd (stochastic subgradient descent) optimize vector w.
the classification methos sign(w*x + bias).
question how find best bias it?
guess has during w optimizing, how? have no idea.
your hypothesis sign(<w, x> + b), think second x' = [x 1], express hypothesis sign(<w', x'>), w' = [w b]. hope shows b not different w's (the difference regularizing term ||w||^2 not involve b). need d l/ d b l loss function.
Comments
Post a Comment