ChoiceNet: Robust Learning by Revealing Output Correlations

Sungjoon Choi, Sanghoon Hong, Kyungjae Lee, Sungbin Lim

Abstract: In this paper, we focus on the supervised learning problem with corrupt training data. We assume that the training dataset is generated from a mixture of a target distribution and other unknown distributions. We estimate the quality of each data by revealing the correlation between the generated distribution and the target distribution. To this end, we present a novel framework referred to here as ChoiceNet that can robustly infer the target distribution in the presence of inconsistent data. We demonstrate that the proposed framework is applicable to both classification and regression tasks. Particularly, ChoiceNet is evaluated in comprehensive experiments, where we show that it constantly outperforms existing baseline methods in the handling of noisy data in synthetic regression tasks as well as behavior cloning problems. In the classification tasks, we apply the proposed method to the MNIST and CIFAR-10 datasets and it shows superior performances in terms of robustness to different types of noisy labels.

Keywords: Robust Deep Learning, weakly supervised learning