DenseForcePosFirstWeight: Create a layer that enforces positive weights in the first...

In enweg/quickbnnr: Quick way to implement Bayesian Neural Networks in R and estimating these using Julia's Turing

Create a layer that enforces positive weights in the first column of the weight matrix. This often solves some identification issues.

Description

Priors for all biases and weights not in the first weight matrix column are standard normal. Priors for the weights in the first column are truncted standard normal (0, inf).

Usage

DenseForcePosFirstWeight(
  in_size,
  out_size,
  activation = c("identity", "sigmoid", "tanh", "relu")
)

Arguments

in_size	input size
out_size	output size
activation	activation function

enweg/quickbnnr documentation built on April 15, 2022, 3:29 a.m.