Fully Connected Neural Networks with Self-Control of Noise Levels

Abstract
We propose a generalization of fully connected neural networks, introducing a nonlinear mechanism of the noise level self-control.

This publication has 24 references indexed in Scilit: