Struct VbHmmResult 

pub struct VbHmmResult {
    pub alpha: Vec<f64>,
    pub beta: Vec<f64>,
    pub gamma: Vec<f64>,
    pub elbo_history: Vec<f64>,
    pub n_iter: usize,
    pub converged: bool,
}

Result of Variational Bayes HMM training.

Fields

alpha: Vec<f64>

Dirichlet concentration parameters for the initial-state posterior (n_states,).

beta: Vec<f64>

Dirichlet concentration parameters for the transition posterior, row-major (n_states × n_states,).

gamma: Vec<f64>

Dirichlet concentration parameters for the emission posterior, row-major (n_states × n_obs,).

elbo_history: Vec<f64>

ELBO (Evidence Lower BOund) at each iteration.

n_iter: usize

Number of VB-EM iterations executed.

converged: bool

Whether the algorithm converged within the tolerance.

Implementations

impl VbHmmResult

pub fn expected_log_pi(&self) -> Vec<f64>

Expected log initial-state probabilities: E[log π_i] = ψ(α_i) − ψ(Σ_j α_j).

pub fn mean_pi(&self) -> Vec<f64>

Posterior mean of the initial-state distribution: α_i / Σ_j α_j.

pub fn mean_a(&self) -> Vec<f64>

Posterior mean of the transition matrix (n_states × n_states, row-major).

pub fn mean_b(&self) -> Vec<f64>

Posterior mean of the emission matrix (n_states × n_obs, row-major).

Trait Implementations

impl Clone for VbHmmResult

fn clone(&self) -> VbHmmResult

Returns a duplicate of the value.

fn clone_from(&mut self, source: &Self)

Performs copy-assignment from source.

impl Debug for VbHmmResult

fn fmt(&self, f: &mut Formatter<'_>) -> Result

Formats the value using the given formatter.