renormLayer.m

From Spinach Documentation Wiki
Revision as of 12:29, 25 April 2026 by Kuprov (talk | contribs) (Normalise sequential blank lines)
Jump to: navigation, search

Renormalisation layer for probability distributions. Makes the mean value across the input vector equal to 1. The mean is used instead of the sum to stay around the sweet spot of single-precision arithmetic.

Syntax

    layer=renormLayer()

Outputs

    layer - Matlab Deep Learning Toolbox layer object

See also

Neural network module

Version 2.8, authors: Ilya Kuprov, Jake Keeley