generate_state: Construct DIVA state list

In ghonk/catlearn.suppls: Catlearn Supplemental Functions

Description

Construct the state list. The primary use of this function is to construct a state list with the default DIVA parameters or generate data-appropriate weights (particularly when the random seed needs to be set).

Usage

generate_state(input, categories, colskip, continuous, make_wts,
  wts_range = NULL, wts_center = NULL, num_hids = NULL,
  learning_rate = NULL, beta_val = NULL, phi = NULL, model_seed = NULL)

Arguments

input	Complete matrix of inputs for training
categories	Vector of category assignment values
colskip	Scalar for number of columns to skip in the tr matrix
continuous	Boolean value indicating if inputs are continuous
make_wts	Boolean value indicating if initial weights should be generated
wts_range	Scalar value for the range of the generated weights
wts_center	Scalar value for the center of the weights
num_hids	Scalar value for the number of hidden units in the model architecture
learning_rate	Learning rate for weight updates through backpropagation
beta_val	Scalar value for the beta parameter
phi	Scalar value for response mapping (Default = 1, meaning no effect)
model_seed	Scalar value to set the random seed

Value

List of the model hyperparameters and weights

ghonk/catlearn.suppls documentation built on May 3, 2019, 5:16 p.m.