Loading...

Memristor crossbar-based hardware implementation of the IDS method

Merrikh Bayat, F ; Sharif University of Technology | 2011

635 Viewed
  1. Type of Document: Article
  2. DOI: 10.1109/TFUZZ.2011.2160024
  3. Publisher: 2011
  4. Abstract:
  5. Ink drop spread (IDS) is the engine of an active learning method, which is the methodology of soft computing. IDS, as a pattern-based processing unit, extracts useful information from a system that is subjected to modeling. In spite of its excellent potential to solve problems such as classification and modeling compared with other soft-computing tools, finding its simple and fast hardware implementation is still a challenge. This paper describes a new hardware implementation of the IDS method that is based on the memristor crossbar structure. In addition to simplicity, being completely real time, having low latency, and the ability to continue working properly after the occurrence of power failure are some of the advantages of our proposed circuit. Moreover, some of operations in the IDS method have fuzzy nature, and as we will show at the end of this paper, updation of rules in the IDS structure and spiky neural networks are very similar. Therefore, IDS can be considered as a new fuzzy implementation of artificial spiky neural networks as well
  6. Keywords:
  7. Active learning method (ALM) ; Memristor crossbar ; Cross-bar structures ; Fuzzy implementation ; Hardware implementations ; ink drop spread (IDS) ; Ink drop spreads ; Low latency ; Power failure ; Processing units ; Real time ; Soft computing tools ; Drops ; Fuzzy inference ; Fuzzy neural networks ; Learning systems ; Memristors ; Passive filters ; Resistors ; Soft computing ; Hardware
  8. Source: IEEE Transactions on Fuzzy Systems ; Volume 19, Issue 6 , Dec , 2011 , Pages 1083-1096 ; 10636706 (ISSN)
  9. URL: http://ieeexplore.ieee.org/xpl/articleDetails.jsp?arnumber=5893932