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 1806.06949cs.LGstat.ML
收藏

基于修剪后的权重预算的全深度神经网络训练

Full deep neural network training on a pruned weight budget

The University of British Columbia(不列颠哥伦比亚大学);Mercedes-Benz Research & Development North America(梅赛德斯-奔驰北美研发中心)

Maximilian Golub, Guy Lemieux, Mieszko Lis

摘要:

我们引入了一种DNN训练技术,该技术只学习完整参数集的一小部分,而不会导致准确性损失。为此,我们的算法将反向传播期间更新的权重总数限制为具有最高总梯度的权重。其余的权重不被跟踪,并且它们的初始值在每次访问时重新生成,以避免将它们存储在内存中。这可以大大减少训练和推理过程中的片外存储器访问次数,这是DNN加速器能源需求的关键组成部分。通过确保总权重扩散保持接近基线未修剪的SGD,使用我们的技术修剪的网络能够在网络架构中保持最先进的准确性-包括以前被确定为难以压缩的网络,如Densenet和WRN。在ImageNet上使用ResNet 18,我们观察到11.7 $\times$的重量减少而没有准确性损失,高达24.4 $\times$,准确性影响很小。

英文摘要:

We introduce a DNN training technique that learns only a fraction of the full parameter set without incurring an accuracy penalty. To do this, our algorithm constrains the total number of weights updated during backpropagation to those with the highest total gradients. The remaining weights are not tracked, and their initial value is regenerated at every access to avoid storing them in memory. This can dramatically reduce the number of off-chip memory accesses during both training and inference, a key component of the energy needs of DNN accelerators. By ensuring that the total weight diffusion remains close to that of baseline unpruned SGD, networks pruned using our technique are able to retain state-of-the-art accuracy across network architectures -- including networks previously identified as difficult to compress, such as Densenet and WRN. With ResNet18 on ImageNet, we observe an 11.7$\times$ weight reduction with no accuracy loss, and up to 24.4$\times$ with a small accuracy impact.

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。

↑