Stochastic Neural Networks with Layer-Wise Adjustable Sequence Length

被引：0

作者：

Wang, Ziheng ^{[1
]}

Reviriego, Pedro ^{[2
]}

Niknia, Farzad ^{[1
]}

Liu, Shanshan ^{[3
]}

Gao, Zhen ^{[4
]}

Lombardi, Fabrizio ^{[1
]}

机构：

[1] Northeastern Univ, Dept Elect & Comp Engn, Boston, MA 02115 USA

[2] Univ Politecn Madrid, ETSI Telecomunicac, Madrid 28040, Spain

[3] Univ Elect Sci & Technol China, Sch Informat & Commun Engn, Chengdu 611731, Peoples R China

[4] Tianjin Univ, Sch Informat & Engn, Tianjin, Peoples R China

来源：

2024 IEEE 24TH INTERNATIONAL CONFERENCE ON NANOTECHNOLOGY, NANO 2024 | 2024年

关键词：

INTERNET; CIRCUITS;

D O I：

10.1109/NANO61778.2024.10628894

中图分类号：

TM [电工技术]; TN [电子技术、通信技术];

学科分类号：

0808 ; 0809 ;

摘要：

The implementation of Neural Networks (NNs) on resource-limited devices poses significant challenges, with Stochastic Computing (SC) as a solution for efficient execution. By representing values as sequences of bits that are processed serially, SC implementations significantly reduce the energy dissipation of NNs. However, with NNs growing in size, SC can also become energy-intensive, prompting a need for enhanced efficiency. This paper introduces Adjustable Sequence Length (ASL), a method employing varied sequence lengths across different NN layers to reduce energy/latency overheads with negligible impact on performance. The feasibility of sequence truncation across layers is assessed; as per simulation results, the ASL method demonstrates significant savings in energy of up to 40.41% and the savings in latency of up to 54.74% when compared with conventional SC implementations.

引用

页码：436 / 441

页数：6