Learning to Reweight with Deep Interactions

TL;DR

This paper introduces a novel data reweighting method where the teacher model leverages the internal states of the student model, leading to improved training performance in image classification and neural machine translation tasks.

Contribution

It proposes an enhanced learning to reweight algorithm that incorporates student internal states into the teacher model, jointly trained with meta gradients for better data weighting.

Findings

Significant improvement over previous methods in image classification.

Effective handling of noisy labels in training.

Enhanced neural machine translation performance.

Abstract

Recently, the concept of teaching has been introduced into machine learning, in which a teacher model is used to guide the training of a student model (which will be used in real tasks) through data selection, loss function design, etc. Learning to reweight, which is a specific kind of teaching that reweights training data using a teacher model, receives much attention due to its simplicity and effectiveness. In existing learning to reweight works, the teacher model only utilizes shallow/surface information such as training iteration number and loss/accuracy of the student model from training/validation sets, but ignores the internal states of the student model, which limits the potential of learning to reweight. In this work, we propose an improved data reweighting algorithm, in which the student model provides its internal states to the teacher model, and the teacher model returns adaptive weights of training samples to enhance the training of the student model. The teacher model is jointly trained with the student model using meta gradients propagated from a validation set. Experiments on image classification with clean/noisy labels and neural machine translation empirically demonstrate that our algorithm makes significant improvement over previous methods.