nn_model_sel: Model selection for a MLP with a single hidden layer.

In andrew-mcinerney/nnic: Statistical-modelling approach to Neural Network

Model selection for a MLP with a single hidden layer.

Description

This function determines the optimal number of hidden units to use in a single-hidden-layer MLP. It allows for a bottom-up and top-down approach.

Usage

nn_model_sel(
  X,
  Y,
  q_max,
  q_min = 1,
  W = NULL,
  n_iter = 1,
  inf_crit = "BIC",
  unif = 3,
  method = "top_down",
  remove = "best",
  plot = F,
  dev = unif/2,
  ...
)

Arguments

X	Matrix of inputs
Y	Output vector
q_max	Largest number of hidden units to consider
n_iter	Number of iterations
inf_crit	Information criterion
unif	Uniform distribution
method	Procedure

Value

Optimal number of hidden units

andrew-mcinerney/nnic documentation built on Dec. 9, 2022, 1:26 p.m.