Advanced Knowledge Transfer: Refined Feature Distillation for Zero-Shot Quantization in Edge Computing

TL;DR

This paper presents AKT, a novel feature distillation method that enhances zero-shot quantization of low-bit models by utilizing spatial and channel attention, significantly improving accuracy on CIFAR datasets.

Contribution

AKT introduces a new knowledge transfer approach that refines feature maps using attention mechanisms, addressing low-bit quantization challenges in ZSQ.

Findings

Achieved state-of-the-art accuracy in 3 and 5-bit models on CIFAR-10.

Effectively addresses gradient explosion in low-bit models.

Demonstrated significant performance improvements over existing methods.

Abstract

We introduce AKT (Advanced Knowledge Transfer), a novel method to enhance the training ability of low-bit quantized (Q) models in the field of zero-shot quantization (ZSQ). Existing research in ZSQ has focused on generating high-quality data from full-precision (FP) models. However, these approaches struggle with reduced learning ability in low-bit quantization due to its limited information capacity. To overcome this limitation, we propose effective training strategy compared to data generation. Particularly, we analyzed that refining feature maps in the feature distillation process is an effective way to transfer knowledge to the Q model. Based on this analysis, AKT efficiently transfer core information from the FP model to the Q model. AKT is the first approach to utilize both spatial and channel attention information in feature distillation in ZSQ. Our method addresses the fundamental gradient exploding problem in low-bit Q models. Experiments on CIFAR-10 and CIFAR-100 datasets demonstrated the effectiveness of the AKT. Our method led to significant performance enhancement in existing generative models. Notably, AKT achieved significant accuracy improvements in low-bit Q models, achieving state-of-the-art in the 3,5bit scenarios on CIFAR-10. The code is available at https://github.com/Inpyo-Hong/AKT-Advanced-knowledge-Transfer.