arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 1611.06342cs.LGcs.NE
收藏

权容量约束下的量化神经网络设计

Quantized neural network design under weight capacity constraint

Seoul National University(首尔大学)

Sungho Shin, Kyuyeon Hwang, Wonyong Sung

摘要:

深度神经网络算法的硬件实现复杂度可以通过缩放单元的数量或减少权重的字长来降低。然而,这两种方法都可能伴随着性能下降,尽管进行了许多类型的研究来缓解这个问题。因此,在网络规模缩放和权值量化之间,哪一个更有效,是一个重要的问题,对于硬件优化。在这项研究中,在改变网络复杂度和权重字长的同时,评估了全连接深度神经网络(FCDNN)和卷积神经网络(CNN)的性能。基于这些实验,我们提出了有效压缩比(ECR),以指导网络规模和权重的精度之间的权衡时,硬件资源是有限的。

英文摘要:

The complexity of deep neural network algorithms for hardware implementation can be lowered either by scaling the number of units or reducing the word-length of weights. Both approaches, however, can accompany the performance degradation although many types of research are conducted to relieve this problem. Thus, it is an important question which one, between the network size scaling and the weight quantization, is more effective for hardware optimization. For this study, the performances of fully-connected deep neural networks (FCDNNs) and convolutional neural networks (CNNs) are evaluated while changing the network complexity and the word-length of weights. Based on these experiments, we present the effective compression ratio (ECR) to guide the trade-off between the network size and the precision of weights when the hardware resource is limited.

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。

↑