English

StableKD: Breaking Inter-block Optimization Entanglement for Stable Knowledge Distillation

Computer Vision and Pattern Recognition 2024-09-24 v2

Abstract

Knowledge distillation (KD) has been recognized as an effective tool to compress and accelerate models. However, current KD approaches generally suffer from an accuracy drop and/or an excruciatingly long distillation process. In this paper, we tackle the issue by first providing a new insight into a phenomenon that we call the Inter-Block Optimization Entanglement (IBOE), which makes the conventional end-to-end KD approaches unstable with noisy gradients. We then propose StableKD, a novel KD framework that breaks the IBOE and achieves more stable optimization. StableKD distinguishes itself through two operations: Decomposition and Recomposition, where the former divides a pair of teacher and student networks into several blocks for separate distillation, and the latter progressively merges them back, evolving towards end-to-end distillation. We conduct extensive experiments on CIFAR100, Imagewoof, and ImageNet datasets with various teacher-student pairs. Compared to other KD approaches, our simple yet effective StableKD greatly boosts the model accuracy by 1% ~ 18%, speeds up the convergence up to 10 times, and outperforms them with only 40% of the training data.

Keywords

Cite

@article{arxiv.2312.13223,
  title  = {StableKD: Breaking Inter-block Optimization Entanglement for Stable Knowledge Distillation},
  author = {Shiu-hong Kao and Jierun Chen and S. H. Gary Chan},
  journal= {arXiv preprint arXiv:2312.13223},
  year   = {2024}
}
R2 v1 2026-06-28T13:57:49.955Z