UWC: Unit-wise Calibration Towards Rapid Network Compression
This work addresses the problem of efficient network compression for deployment in resource-constrained environments, representing an incremental improvement over existing post-training quantization methods.
The paper tackles the challenge of maintaining high performance in Convolutional Neural Networks during post-training quantization to very low bit-widths (e.g., INT4 and INT3) by proposing a unit-wise feature reconstruction algorithm that leverages interactions between adjacent layers to minimize quantization error, achieving near-original accuracy on ImageNet and COCO datasets.
This paper introduces a post-training quantization~(PTQ) method achieving highly efficient Convolutional Neural Network~ (CNN) quantization with high performance. Previous PTQ methods usually reduce compression error via performing layer-by-layer parameters calibration. However, with lower representational ability of extremely compressed parameters (e.g., the bit-width goes less than 4), it is hard to eliminate all the layer-wise errors. This work addresses this issue via proposing a unit-wise feature reconstruction algorithm based on an observation of second order Taylor series expansion of the unit-wise error. It indicates that leveraging the interaction between adjacent layers' parameters could compensate layer-wise errors better. In this paper, we define several adjacent layers as a Basic-Unit, and present a unit-wise post-training algorithm which can minimize quantization error. This method achieves near-original accuracy on ImageNet and COCO when quantizing FP32 models to INT4 and INT3.