covariance: Indices for covariances in the neural network

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

Description

This function assigns indices for all covariance elements in the neural network.

Usage

covariance(NN)

Arguments

NN	Lists the structure of the neural network

Value

NN with new elements:

• Indices (weights and activation units) for deterministic matrix F * μ_{WA} for each layer

• Bias indices for deterministic matrix F * μ_{B} for each layer

• Indices (weights and activation units) for deterministic matrix F * Σ_{ZWA} for each layer

• Indices for the parameter update step for each layer

• Indices for the hidden state update step for each layer

• Indices (weights and activation units) for deterministic matrix F * Σ_{WAθ} for each layer

• Indices for activation unit for each layer

• Bias indices for deterministic matrix F * Σ_{B} for each layer

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