Trait potpourri::mixture::Latent

pub trait Latent<T>where
    T: Parametrizable,{
    // Required method
    fn expect(
        &self,
        data: &T::DataIn<'_>,
        likelihood: &T::Likelihood
    ) -> Result<(T::Likelihood, AvgLLH), Error>;
}

An additional interface for Mixables that can be used as latent states. These can be categorical distributions, with or without finite Dirichlet or infinite Dirichlet process priors. The Mixables are here used not multiple components but only as one distribution of the latent states.

Required Methods

fn expect( &self, data: &T::DataIn<'_>, likelihood: &T::Likelihood ) -> Result<(T::Likelihood, AvgLLH), Error>

Implementors

impl Latent<Finite> for Finite

FIXME compute the likelihood.. is it just the sum of all?