Online Multiclass Classification Based on Prediction Margin for Partial Feedback
Abstract
We consider the problem of online multiclass classification with partial feedback, where an algorithm predicts a class for a new instance in each round and only receives its correctness. Although several methods have been developed for this problem, recent challenging real-world applications require further performance improvement. In this paper, we propose a novel online learning algorithm inspired by recent work on learning from complementary labels, where a complementary label indicates a class to which an instance does not belong. This allows us to handle partial feedback deterministically in a margin-based way, where the prediction margin has been recognized as a key to superior empirical performance. We provide a theoretical guarantee based on a cumulative loss bound and experimentally demonstrate that our method outperforms existing methods which are non-margin-based and stochastic.
Cite
@article{arxiv.1902.01056,
title = {Online Multiclass Classification Based on Prediction Margin for Partial Feedback},
author = {Takuo Kaneko and Issei Sato and Masashi Sugiyama},
journal= {arXiv preprint arXiv:1902.01056},
year = {2019}
}