Struct MultiClass 

pub struct MultiClass { /* private fields */ }

Multi-class Tsetlin Machine using one-vs-all strategy.

Each class has its own set of clauses. Prediction is the class with the highest vote sum.

Example

use tsetlin_rs::{Config, MultiClass};

let config = Config::builder().clauses(100).features(4).build().unwrap();
let mut tm = MultiClass::new(config, 3, 50);

// Train on data where label is class index (0, 1, or 2)
let x = vec![vec![1, 1, 0, 0], vec![0, 0, 1, 1]];
let y = vec![0, 1];
tm.fit(&x, &y, 100, 42);

Implementations

impl MultiClass

pub fn new(config: Config, n_classes: usize, threshold: i32) -> Self

Creates multi-class machine with given number of classes.

Arguments

• config - Machine configuration (clauses, features, etc.)
• n_classes - Number of output classes
• threshold - Vote threshold for training

pub fn n_classes(&self) -> usize

Returns the number of classes.

pub fn class_votes(&self, x: &[u8]) -> Vec<f32>

Computes vote sums for each class.

pub fn predict(&self, x: &[u8]) -> usize

Predicts class with highest vote sum.

Returns 0 if votes are empty or contain NaN.

pub fn train_one<R: Rng>(&mut self, x: &[u8], y: usize, rng: &mut R)

Trains on a single example.

pub fn fit(&mut self, x: &[Vec<u8>], y: &[usize], epochs: usize, seed: u64)

Trains for given number of epochs.

Arguments

• x - Training inputs (binary features)
• y - Class labels (0 to n_classes-1)
• epochs - Number of training epochs
• seed - Random seed for reproducibility

pub fn evaluate(&self, x: &[Vec<u8>], y: &[usize]) -> f32

Evaluates classification accuracy on test data.

Returns fraction of correct predictions (0.0 to 1.0).

pub fn quick( n_clauses: usize, n_features: usize, n_classes: usize, threshold: i32, ) -> Self

Quick constructor with sensible defaults.

Panics

Panics if n_clauses is odd or zero, or n_features is zero.

pub fn partial_fit(&mut self, x: &[u8], y: usize, seed: u64)

Trains on a single sample (online/incremental learning).

Arguments

• x - Single input sample (binary features)
• y - Class label (0 to n_classes-1)
• seed - Random seed for this update

pub fn partial_fit_batch(&mut self, xs: &[Vec<u8>], ys: &[usize], seed: u64)

Trains on a mini-batch of samples (online/incremental learning).

Arguments

• xs - Batch of input samples
• ys - Batch of class labels
• seed - Random seed for this batch

Trait Implementations

impl Clone for MultiClass

fn clone(&self) -> MultiClass

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl Debug for MultiClass

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

Formats the value using the given formatter.

impl TsetlinModel<Vec<u8>, usize> for MultiClass

fn fit(&mut self, x: &[Vec<u8>], y: &[usize], epochs: usize, seed: u64)

Trains the model on labeled data.

fn predict(&self, x: &Vec<u8>) -> usize

Predicts label for a single sample.

fn evaluate(&self, x: &[Vec<u8>], y: &[usize]) -> f32

Evaluates model accuracy/performance on test data.

fn predict_batch(&self, xs: &[X]) -> Vec<Y>

Batch prediction for multiple samples.