%Aigaion2 BibTeX export from Idiap Publications
%Thursday 21 November 2024 12:40:35 PM

@ARTICLE{Moerland-96.4,
         author = {Moerland, Perry and Fiesler, Emile and Saxena, Indu},
       keywords = {(artificial) neural network, activation function, curve fit, gain, hardware implementation, liquid crystal light valve (LCLV), optical multilayer neural network},
       projects = {Idiap},
          title = {Incorporation of Liquid-Crystal Light Valve Non-Linearities in Optical Multilayer Neural Networks},
        journal = {Applied Optics},
         volume = {35},
         number = {26},
           year = {1996},
      publisher = {The Optical Society of America (OSA)},
        address = {New York, New York},
           issn = {0003-6935},
       abstract = {Sigmoidlike activation functions, as available in analog hardware, differ in various ways from the standard sigmoidal function because they are usually asymmetric, truncated, and have a non-standard gain. We present an adaptation of the backpropagation learning rule to compensate for these nonstandard sigmoids. This method is applied to multilayer neural networks with all-optical forward propagation and liquid-crystal light valves (LCLV) as optical thresholding devices. In this paper, the results of simulations of a backpropagation neural network with five different LCLV response curves as activation functions are presented. Although LCLV's perform poorly with the standard backpropagation algorithm, it is shown that our adapted learning rule performs well with these LCLV curves.},
            pdf = {https://publications.idiap.ch/attachments/reports/1996/LCLV.pdf},
ipdmembership={learning},
language={English},
}