Skip to contents

Draw synthetic base rows from a trained texture model

Source: R/gan.R
sample_texture.Rd

Binary columns (detected at training time) are rounded to \({0, 1}\) after de-normalisation, and one-hot dummy groups are made mutually exclusive (the column with the highest pre-rounding value wins per row).

Usage

sample_texture(trained_gan, n)

Arguments

trained_gan

An iconic_gan object from train_gan_on_real_data().

n

Number of rows to draw.

Value

A data frame of n rows with the trained columns, on the original (de-normalised) scale. Binary columns contain only 0/1 values.

Examples

if (check_torch_setup()) {
  dat <- load_real_input_data(example = TRUE)
  gan <- train_gan_on_real_data(dat$gan_training_data, epochs = 5,
    verbose = FALSE)
  head(sample_texture(gan, 5))
}