Efficient Interactive Multiclass Learning from Binary Feedback

Ngo, Hung; Luciw, Matthew; Nagi, Jawas; Forster, Alexander; Schmidhuber, Jürgen; Vien, Ngo Anh

doi:10.1145/2629631

Public

Efficient Interactive Multiclass Learning from Binary Feedback

Published in ACM Transactions on Interactive Intelligent Systems • Aug 11, 2014

NobleIDNI3P16W80R17S47

Authors:

Hung Ngo

,

Matthew Luciw

,

Jawas Nagi

Abstract

We introduce a novel algorithm called upper confidence - weighted learning (UCWL) for online multiclass learning from binary feedback (e.g., feedback that indicates whether the prediction was right or wrong). UCWL combines the upper confidence bound (UCB) framework with the soft confidence-weighted ...

Discussions