Answers are grounded in this paper and cite their sources.
Investigates subgroup performance improvements in knowledge distillation, indicating biases can be mitigated.
Knowledge distillation is widely used as a means of improving the performance of a relatively simple student model using the predictions from a complex teacher model. Several works have shown that distillation significantly boosts the student's overall performance; however, are these gains uniform across all data subgroups? In this paper, we show that distillation can harm performance on certain subgroups, e.g., classes with few associated samples. We trace this behaviour to errors made by the teacher distribution being transferred to and amplified by the student model. To mitigate this problem, we present techniques which soften the teacher influence for subgroups where it is less reliable. Experiments on several image classification benchmarks show that these modifications of distillation maintain boost in overall accuracy, while additionally ensuring improvement in subgroup performance.
No takes yet. Share an insight, caveat, or question.
Łukasik et al. (2021) studied this question.
Synapse has enriched 8 closely related papers on similar clinical questions. Consider them for comparative context: