Description This repository contains an experimental PyTorch implementation exploring the NoProp algorithm, presented in the paper "NOPROP: TRAINING NEURAL NETWORKS WITHOUT BACK-PROPAGATION OR FORWARD ...

A nonsynaptic, backpropagation-based learning algorithm powered by stochastic gradient descent is put forward to iteratively optimize four parameters of the generalized sigmoid transfer function ...

Classification is one-out-of several applications in the neural network (NN) world. Multilayer perceptron (MLP) is the common neural network architecture which ...