English

Channel Pruning In Quantization-aware Training: An Adaptive Projection-gradient Descent-shrinkage-splitting Method

Machine Learning 2022-04-12 v1 Artificial Intelligence Computer Vision and Pattern Recognition Numerical Analysis Numerical Analysis

Abstract

We propose an adaptive projection-gradient descent-shrinkage-splitting method (APGDSSM) to integrate penalty based channel pruning into quantization-aware training (QAT). APGDSSM concurrently searches weights in both the quantized subspace and the sparse subspace. APGDSSM uses shrinkage operator and a splitting technique to create sparse weights, as well as the Group Lasso penalty to push the weight sparsity into channel sparsity. In addition, we propose a novel complementary transformed l1 penalty to stabilize the training for extreme compression.

Cite

@article{arxiv.2204.04375,
  title  = {Channel Pruning In Quantization-aware Training: An Adaptive Projection-gradient Descent-shrinkage-splitting Method},
  author = {Zhijian Li and Jack Xin},
  journal= {arXiv preprint arXiv:2204.04375},
  year   = {2022}
}
R2 v1 2026-06-24T10:43:02.929Z