Dynamic selection for reconstructing instance-dependent noisy labels

Jie Yang, Xiaoguang Niu, Yuanzhuo Xu, Zejun Zhang, Guangyi Guo, Steve Drew, Ruizhi Chen

Abstract: Highlights•We propose a dynamic sample selection method when performing label reconstruction.•The proposed method is adaptive to different noise rates.•Instance features are used along with classifier outputs to reconstruct noisy labels.