arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~
arXiv 2203.02549cs.CV
收藏

结构化修剪是修剪CNN所需的全部

Structured Pruning is All You Need for Pruning CNNs at Initialization

Cornell University(康奈尔大学)

Yaohui Cai, Weizhe Hua, Hongzheng Chen, G. Edward Suh, Christopher De Sa, Zhiru Zhang

摘要:

剪枝是一种用于减小卷积神经网络(CNN)的模型大小和计算成本的流行技术。然而,通常需要缓慢的重新训练或微调过程来恢复由修剪引起的准确性损失。最近,提出了一种关于权重修剪的新研究方向,即初始化修剪(PAI),在训练之前直接修剪CNN,从而可以避免微调或重新训练。虽然PAI在减少模型大小方面显示出有希望的结果,但现有的方法依赖于细粒度的权重修剪,这需要非结构化的稀疏矩阵计算,除非稀疏度非常高,否则难以在实践中实现真正的加速。这项工作是第一次表明,细粒度的重量修剪实际上是没有必要的PAI。相反,逐层压缩比是确定在初始化时修剪的CNN模型的准确性的主要关键因素。基于这一关键的观察,我们提出了预裁剪,结构化的硬件高效的模型压缩计划。预裁剪按照逐层压缩比在通道级别直接压缩模型。与权值修剪相比,该方案在不牺牲精度的前提下,在存储和计算上都具有规则性和密集性。此外,由于PreCropping在初始化时压缩了CNN,因此在商品硬件上进行训练和推理时,CNN的计算和内存成本都有所降低。我们在几个现代CNN架构上实证了我们的方法,包括CIFAR-10和ImageNet的ResNet,ShuffleNet和MobileNet。

英文摘要:

Pruning is a popular technique for reducing the model size and computational cost of convolutional neural networks (CNNs). However, a slow retraining or fine-tuning procedure is often required to recover the accuracy loss caused by pruning. Recently, a new research direction on weight pruning, pruning-at-initialization (PAI), is proposed to directly prune CNNs before training so that fine-tuning or retraining can be avoided. While PAI has shown promising results in reducing the model size, existing approaches rely on fine-grained weight pruning which requires unstructured sparse matrix computation, making it difficult to achieve real speedup in practice unless the sparsity is very high. This work is the first to show that fine-grained weight pruning is in fact not necessary for PAI. Instead, the layerwise compression ratio is the main critical factor to determine the accuracy of a CNN model pruned at initialization. Based on this key observation, we propose PreCropping, a structured hardware-efficient model compression scheme. PreCropping directly compresses the model at the channel level following the layerwise compression ratio. Compared to weight pruning, the proposed scheme is regular and dense in both storage and computation without sacrificing accuracy. In addition, since PreCropping compresses CNNs at initialization, the computational and memory costs of CNNs are reduced for both training and inference on commodity hardware. We empirically demonstrate our approaches on several modern CNN architectures, including ResNet, ShuffleNet, and MobileNet for both CIFAR-10 and ImageNet.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑