Memristor Crossbar-Based Neuromorphic Computing System: A Case Study

• Published in: IEEE transaction on neural networks and learning systems Vol. 25; no. 10; pp. 1864 - 1878
• Main Authors: Miao Hu, Hai Li, Yiran Chen, Qing Wu, Rose, Garrett S., Linderman, Richard W.
• Format: Journal Article
• Language: English
• Published: United States IEEE 01.10.2014; The Institute of Electrical and Electronics Engineers, Inc. (IEEE)
• Subjects: Arrays; Associative memory; Biological; Biological neural networks; brain-state-in-a-box (BSB); Computer simulation; Computer Storage Devices; Computer Systems; Computers; crossbar array; Hardware; Humans; Mathematical analysis; Mathematical models; Memory devices; memristor; Memristors; Monte Carlo simulation; Nanotechnology - instrumentation; Nanotechnology - methods; Neural networks; Neural Networks (Computer); neuromorphic hardware; Neuromorphics; Neurons; Resistors; Training; neuromorphic hardware; Associative memory; brain-state-in-a-box (BSB); crossbar array; memristor
• ISSN: 2162-237X; 2162-2388; 2162-2388
• DOI: 10.1109/TNNLS.2013.2296777

By mimicking the highly parallel biological systems, neuromorphic hardware provides the capability of information processing within a compact and energy-efficient platform. However, traditional Von Neumann architecture and the limited signal connections have severely constrained the scalability and performance of such hardware implementations. Recently, many research efforts have been investigated in utilizing the latest discovered memristors in neuromorphic systems due to the similarity of memristors to biological synapses. In this paper, we explore the potential of a memristor crossbar array that functions as an autoassociative memory and apply it to brain-state-in-a-box (BSB) neural networks. Especially, the recall and training functions of a multianswer character recognition process based on the BSB model are studied. The robustness of the BSB circuit is analyzed and evaluated based on extensive Monte Carlo simulations, considering input defects, process variations, and electrical fluctuations. The results show that the hardware-based training scheme proposed in the paper can alleviate and even cancel out the majority of the noise issue.