buildCzz: Reformat covariance matrix between units of the previous and...

In mgoulet847/tagi: Tractable Approximate Gaussian Inference in Neural Networks

Description

This function properly reformats covariance matrix between units of the previous and current layers Σ_{ZZ^{+}} for the update step.

Usage

buildCzz(Czz, currentHiddenUnit, prevHiddenUnit, batchSize)

Arguments

Czz	Covariance matrix between units of the previous and current layers
currentHiddenUnit	Number of units in the current layer
prevHiddenUnit	Number of units in the previous layer
batchSize	Number of observations trained at the same time

Value

Reformatted covariance matrix between of the previous and current layers

mgoulet847/tagi documentation built on Dec. 21, 2021, 5:10 p.m.