The artificial neural network (ANN) has shown the effectiveness in unmanned systems and many other fields. Training an ANN with CPUs or other digital circuits can take a long time. And thus some analog neural network models have been composed to utilize the inherent advantage of analog circuits. However there are some problems in these models when considering about the flexibility and the speed. In this paper, we use op-amp integrators to store the weights of an analog ANN, and use the feedback of the circuit to train the network. It is easy to hold or change the weights in our circuit and the training of a sample can be automatically done quickly. The circuit only supports the ANN with one output in this paper. Besides more possible utilizations of the circuit are also proposed.


    Access

    Check access

    Check availability in my library

    Order at Subito €


    Export, share and cite



    Title :

    Training an Artificial Neural Network with Op-amp Integrators Based Analog Circuits


    Contributors:
    Liu, Weiliang (author) / Li, Zhan (author) / Xue, Shengri (author) / Yang, Xuebo (author) / Lin, Weiyang (author)


    Publication date :

    2018-08-01


    Size :

    2431211 byte




    Type of media :

    Conference paper


    Type of material :

    Electronic Resource


    Language :

    English



    Discrete-time state estimation of analog double integrators

    Shats, S. / Bobrovsky, B.Z. / Shaked, U. | IEEE | 1988


    Method for training artificial neural network

    HASBERG CARSTEN / NASR TOMER / SARANRITICHAI PARTHA | European Patent Office | 2022

    Free access

    Contractor/Integrators

    Online Contents | 2011


    Contractors/Integrators

    Online Contents | 2012


    Contractor/Integrators

    Online Contents | 2010