Synchronous Pipeline Circuit Design for an Adaptive Neuro-fuzzy Network

Autor:	Ting-Yu Chen, Chun-Chang Yu, Che-Wei Lin, Jeen-Shing Wang
Rok vydání:	2007
Předmět:	business.industry Computer science Asynchronous communication Circuit design Pipeline (computing) Embedded system Multiplier (economics) Physical design Field-programmable gate array business Circuit extraction Computer hardware Register-transfer level
Zdroj:	Advanced Intelligent Computing Theories and Applications. With Aspects of Artificial Intelligence ISBN: 9783540742012 ICIC (2)
DOI:	10.1007/978-3-540-74205-0_19
Popis:	This paper presents an efficient synchronous pipeline hardware implementation procedure for a neuro-fuzzy (NF) circuit. We decompose the NF circuit into a feedforward circuit and a backpropagation circuit. The concept of pre-calculation to share computation results between the feedforward circuit and backpropagation circuit is introduced to achieve a high throughput rate and low resource usage. A novel pipeline architecture has been adopted to fulfill the concept of pre-calculation. With the unique pipeline architecture, we have successfully enhanced the throughput rate and resource sharing between modules. Particularly, the multiplier usage has been reduced from 7 to 3 and the divider usage from 3 to 1. Finally, we have implemented the NF circuit on FGPA. Our experimental results show a superior performance than that of an asynchronous pipeline design approach and the NF system implemented on MATLAB®.
Databáze:	OpenAIRE
Externí odkaz:	https://explore.openaire.eu/search/publication?articleId=doi_________::e784678d1825660a33a7cfaa29eb2039 https://doi.org/10.1007/978-3-540-74205-0_19 Zobrazit plný text záznamu