Learning Multiclass Classifier Under Noisy Bandit Feedback
This addresses the problem of robust learning under corrupted feedback for machine learning practitioners, representing an incremental improvement with specific theoretical guarantees.
The paper tackles multiclass classification with noisy bandit feedback, where feedback may be flipped with non-zero probability, and proposes a novel approach using unbiased estimators and noise rate estimation, achieving mistake bounds of O(sqrt(T)) in high noise and O(T^(2/3)) in worst-case scenarios.
This paper addresses the problem of multiclass classification with corrupted or noisy bandit feedback. In this setting, the learner may not receive true feedback. Instead, it receives feedback that has been flipped with some non-zero probability. We propose a novel approach to deal with noisy bandit feedback based on the unbiased estimator technique. We further offer a method that can efficiently estimate the noise rates, thus providing an end-to-end framework. The proposed algorithm enjoys a mistake bound of the order of $O(\sqrt{T})$ in the high noise case and of the order of $O(T^{\nicefrac{2}{3}})$ in the worst case. We show our approach's effectiveness using extensive experiments on several benchmark datasets.