Abstract
This article presents a promising new gradient-based backpropagation algorithm for multi-layer
feedforward networks. The method requires no manual selection of global hyperparameters and is capable
of dynamic local adaptations using only first-order information at a low computational cost. Its semistochastic nature makes it fit for mini-batch training and robust to different architecture choices and data
distributions. Experimental evidence shows that the proposed algorithm improves training in terms of both
convergence rate and speed as compared with other well known techniques.
Users
Please
log in to take part in the discussion (add own reviews or comments).