HAWQ: Hessian AWare Quantization of Neural Networks With Mixed-Precision

Zhen Dong, Zhewei Yao, Amir Gholami, Michael W. Mahoney, Kurt Keutzer; Proceedings of the IEEE/CVF International Conference on Computer Vision (ICCV), 2019, pp. 293-302

Abstract


Model size and inference speed/power have become a major challenge in the deployment of neural networks for many applications. A promising approach to address these problems is quantization. However, uniformly quantizing a model to ultra-low precision leads to significant accuracy degradation. A novel solution for this is to use mixed-precision quantization, as some parts of the network may allow lower precision as compared to other layers. However, there is no systematic way to determine the precision of different layers. A brute force approach is not feasible for deep networks, as the search space for mixed-precision is exponential in the number of layers. Another challenge is a similar factorial complexity for determining block-wise fine-tuning order when quantizing the model to a target precision. Here, we introduce Hessian AWare Quantization (HAWQ), a novel second-order quantization method to address these problems. HAWQ allows for the automatic selection of the relative quantization precision of each layer, based on the layer's Hessian spectrum. Moreover, HAWQ provides a deterministic fine-tuning order for quantizing layers. We show the results of our method on Cifar-10 using ResNet20, and on ImageNet using Inception-V3, ResNet50 and SqueezeNext models. Comparing HAWQ with state-of-the-art shows that we can achieve similar/better accuracy with 8x activation compression ratio on ResNet20, as compared to DNAS, and up to 1% higher accuracy with up to 14% smaller models on ResNet50 and Inception-V3, compared to recently proposed methods of RVQuant and HAQ. Furthermore, we show that we can quantize SqueezeNext to just 1MB model size while achieving above 68% top1 accuracy on ImageNet.

Related Material


[pdf] [supp]
[bibtex]
@InProceedings{Dong_2019_ICCV,
author = {Dong, Zhen and Yao, Zhewei and Gholami, Amir and Mahoney, Michael W. and Keutzer, Kurt},
title = {HAWQ: Hessian AWare Quantization of Neural Networks With Mixed-Precision},
booktitle = {Proceedings of the IEEE/CVF International Conference on Computer Vision (ICCV)},
month = {October},
year = {2019}
}