Quick noise-tolerant learning in a multi-layer memristive neural network

abstract

• In this letter, a multi-layer memristive neural network in which memristors serve as memory factor and impact factor is built for the binary images learning. Unlike the traditional artificial neural networks, the memristive neural network is trained with noise samples for the introduction of image overlay. A similarity recognition based on the impact factor for binary images is employed to exhibit the function of the memristive neural network. The Simulinks and the experiments shows that the proposed memristive neural network has quick learning speed and high noise tolerance. 2013 Elsevier B.V.

authors

• Huang, Tingwen

published proceedings

• NEUROCOMPUTING

author list (cited authors)

• Chen, L., Li, C., Huang, T., & Wang, X.

citation count

• 11

complete list of authors

• Chen, Ling||Li, Chuandong||Huang, Tingwen||Wang, Xin

publication date

• April 2014

publisher

• Elsevier  Publisher

published in

• Neurocomputing  Journal

keywords

• Binary
• Image Overlay
• Impact Factor
• Memristive Neural Networks
• Noise

Digital Object Identifier (DOI)

• 10.1016/j.neucom.2013.05.043

start page

• 122

end page

• 126

volume

• 129

URL

• http://dx.doi.org/10.1016/j.neucom.2013.05.043