Rethinking Centered Kernel Alignment in Knowledge Distillation

TL;DR

This paper reexamines the theoretical foundations of Centered Kernel Alignment (CKA) in knowledge distillation, proposing a new framework that simplifies and enhances its application for better performance in model compression tasks.

Contribution

It provides a theoretical analysis of CKA, introduces the RCKA framework linking CKA with MMD, and develops a task-adaptive method that improves distillation efficiency and effectiveness.

Findings

Achieves state-of-the-art results on CIFAR-100, ImageNet-1k, and MS-COCO.

Demonstrates that RCKA reduces computational costs while maintaining performance.

Validates the theoretical connection between CKA and MMD in knowledge distillation.

Abstract

Knowledge distillation has emerged as a highly effective method for bridging the representation discrepancy between large-scale models and lightweight models. Prevalent approaches involve leveraging appropriate metrics to minimize the divergence or distance between the knowledge extracted from the teacher model and the knowledge learned by the student model. Centered Kernel Alignment (CKA) is widely used to measure representation similarity and has been applied in several knowledge distillation methods. However, these methods are complex and fail to uncover the essence of CKA, thus not answering the question of how to use CKA to achieve simple and effective distillation properly. This paper first provides a theoretical perspective to illustrate the effectiveness of CKA, which decouples CKA to the upper bound of Maximum Mean Discrepancy~(MMD) and a constant term. Drawing from this, we propose a novel Relation-Centered Kernel Alignment~(RCKA) framework, which practically establishes a connection between CKA and MMD. Furthermore, we dynamically customize the application of CKA based on the characteristics of each task, with less computational source yet comparable performance than the previous methods. The extensive experiments on the CIFAR-100, ImageNet-1k, and MS-COCO demonstrate that our method achieves state-of-the-art performance on almost all teacher-student pairs for image classification and object detection, validating the effectiveness of our approaches. Our code is available in https://github.com/Klayand/PCKA