Abstract
Equilibrium Propagation (EP) is a novel biologically plausible algorithm for training deep neural networks. However, discrete time implementations of EP could not exploit the full potential of its proposed framework, because the relaxation step is slow on digital architectures. Here, we propose an analog circuit implementation for the relaxation process to accelerate its convergence. In this implementation, the optimization process is executed based on the continuous-time dynamics of EP. Our circuit is validated using a Continuous Hopfield Network prototype emulating the XOR function. This implementation improved the convergence time of EP by a factor of 250 compared to its Python counterpart. As this implementation is scalable to learn more complex functions, it has the potential to be applied on higher-dimensionality datasets.
Sign-in/Register to access full text options 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a callDisclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.
