[−][src]Function bhtsne::run

pub fn run(
    x: &mut [f64], 
    n: usize, 
    d: usize, 
    y: &mut [f64], 
    no_dims: usize, 
    perplexity: f64, 
    theta: f64, 
    skip_random_init: bool, 
    max_iter: u64, 
    stop_lying_iter: u64, 
    mom_switch_iter: u64
)

Performs t-SNE.

Arguments

  • x - A &mut [f64] containing the data to embed.

  • n- A usize that represents the number of d-dimensional points in x.

  • d - A usize that represents the dimension of each point.

  • y - A &mut [f64] where to store the resultant embedding.

  • no_dims - A usize that represents the dimension of the embedded points.

  • perplexity - An f64 representing perplexity value.

  • theta - An f64 representing the theta value. If theta is set to 0.0 the exact version of t-SNE will be used, if set to greater values the Barnes-Hut version will be used instead.

  • skip_random_init - A bool stating wether to skip the random initialization of y. In most cases it should be set to false.

  • max_iter - A u64 indicating the maximum number of iterations to perform the fitting.

  • stop_lying_iter - A u64 indicating the number of the iteration after which the P distribution values become “true“.

  • mom_switch_iter - A u64 indicating the number of the iteration after which the momentum's value is set to its final value.