Obtaining the gradient of what's known as the loss function is an essential step to establish the backpropagation algorithm developed by University of Michigan researchers to train a material ...

The Forward-Forward algorithm (FF) is comparable in speed to backpropagation but has the advantage that it can be used when the precise details of the forward computation are unknown.

A new technical paper titled “The backpropagation algorithm implemented on spiking neuromorphic hardware” was published by University of Zurich, ETH Zurich, Los Alamos National Laboratory, Royal ...

Here, we propose a hardware implementation of the backpropagation algorithm that progressively updates each layer using in situ stochastic gradient descent, avoiding this storage requirement.