Deep geometric knowledge distillation with graphs

TL;DR

This paper introduces a graph-based relative knowledge distillation method that effectively transfers geometric information between neural networks of different sizes, improving accuracy under resource constraints.

Contribution

It proposes a novel graph-based RKD approach that captures latent space geometry, enabling dimension-agnostic knowledge transfer in deep learning models.

Findings

Outperforms existing RKD methods on vision benchmarks

Enables effective knowledge distillation across different network sizes

Improves accuracy for the same computational budget

Abstract

In most cases deep learning architectures are trained disregarding the amount of operations and energy consumption. However, some applications, like embedded systems, can be resource-constrained during inference. A popular approach to reduce the size of a deep learning architecture consists in distilling knowledge from a bigger network (teacher) to a smaller one (student). Directly training the student to mimic the teacher representation can be effective, but it requires that both share the same latent space dimensions. In this work, we focus instead on relative knowledge distillation (RKD), which considers the geometry of the respective latent spaces, allowing for dimension-agnostic transfer of knowledge. Specifically we introduce a graph-based RKD method, in which graphs are used to capture the geometry of latent spaces. Using classical computer vision benchmarks, we demonstrate the ability of the proposed method to efficiently distillate knowledge from the teacher to the student, leading to better accuracy for the same budget as compared to existing RKD alternatives.