Kernel Quantization for Efficient Network Compression
This work addresses efficient network compression for deep learning applications, offering a novel method that improves compression ratios with little performance degradation, though it is incremental in building upon prior pruning and quantization techniques.
This paper tackles the problem of compressing convolutional neural networks by introducing Kernel Quantization (KQ), a framework that quantizes convolution kernels as units to reduce bit-length, achieving state-of-the-art compression ratios with minimal accuracy loss, such as 1.05 bits per parameter on VGG and 1.62 bits on ResNet18 on ImageNet.
This paper presents a novel network compression framework Kernel Quantization (KQ), targeting to efficiently convert any pre-trained full-precision convolutional neural network (CNN) model into a low-precision version without significant performance loss. Unlike existing methods struggling with weight bit-length, KQ has the potential in improving the compression ratio by considering the convolution kernel as the quantization unit. Inspired by the evolution from weight pruning to filter pruning, we propose to quantize in both kernel and weight level. Instead of representing each weight parameter with a low-bit index, we learn a kernel codebook and replace all kernels in the convolution layer with corresponding low-bit indexes. Thus, KQ can represent the weight tensor in the convolution layer with low-bit indexes and a kernel codebook with limited size, which enables KQ to achieve significant compression ratio. Then, we conduct a 6-bit parameter quantization on the kernel codebook to further reduce redundancy. Extensive experiments on the ImageNet classification task prove that KQ needs 1.05 and 1.62 bits on average in VGG and ResNet18, respectively, to represent each parameter in the convolution layer and achieves the state-of-the-art compression ratio with little accuracy loss.