In this work, we propose a Generalized Cross-Entropy-based framework using Chained Deep Learning, termed GCECDL, to code each annotator’s non-stationary patterns regarding the input space while preserving the inter-dependencies among experts through a noise-robust loss function to deal with noisy labels, and a chained deep learning approach. Besides, network self-regularization is achieved by estimating each labeler’s reliability within the chained approach.
Please, if you use this code, cite this paper: Chained Deep Learning Using Generalized Cross-Entropy for Multiple Annotators Classification