English

Bandit Multiclass Linear Classification for the Group Linear Separable Case

Machine Learning 2021-09-01 v2 Machine Learning

Abstract

We consider the online multiclass linear classification under the bandit feedback setting. Beygelzimer, P\'{a}l, Sz\"{o}r\'{e}nyi, Thiruvenkatachari, Wei, and Zhang [ICML'19] considered two notions of linear separability, weak and strong linear separability. When examples are strongly linearly separable with margin γ\gamma, they presented an algorithm based on Multiclass Perceptron with mistake bound O(K/γ2)O(K/\gamma^2), where KK is the number of classes. They employed rational kernel to deal with examples under the weakly linearly separable condition, and obtained the mistake bound of min(K2O~(Klog2(1/γ)),K2O~(1/γlogK))\min(K\cdot 2^{\tilde{O}(K\log^2(1/\gamma))},K\cdot 2^{\tilde{O}(\sqrt{1/\gamma}\log K)}). In this paper, we refine the notion of weak linear separability to support the notion of class grouping, called group weak linear separable condition. This situation may arise from the fact that class structures contain inherent grouping. We show that under this condition, we can also use the rational kernel and obtain the mistake bound of K2O~(1/γlogL))K\cdot 2^{\tilde{O}(\sqrt{1/\gamma}\log L)}), where LKL\leq K represents the number of groups.

Keywords

Cite

@article{arxiv.1912.10340,
  title  = {Bandit Multiclass Linear Classification for the Group Linear Separable Case},
  author = {Jittat Fakcharoenphol and Chayutpong Prompak},
  journal= {arXiv preprint arXiv:1912.10340},
  year   = {2021}
}

Comments

This work is first published in iSAI-NLP 2019, Chiang Mai, Thailand. This version contains larger experiments and fixes calculation errors

R2 v1 2026-06-23T12:53:33.159Z