Struct SparseTsetlinMachine 

pub struct SparseTsetlinMachine { /* private fields */ }

Sparse Tsetlin Machine for memory-efficient inference.

Inference-only model with 5-100x memory reduction compared to dense TsetlinMachine. Create via TsetlinMachine::to_sparse() after training.

Example

use tsetlin_rs::{Config, TsetlinMachine};

let config = Config::builder().clauses(20).features(2).build().unwrap();
let mut tm = TsetlinMachine::new(config, 10);

let x = vec![vec![0, 0], vec![0, 1], vec![1, 0], vec![1, 1]];
let y = vec![0, 1, 1, 0];

tm.fit(&x, &y, 200, 42);

// Convert to sparse for deployment
let sparse = tm.to_sparse();

// Same predictions, less memory
assert_eq!(tm.predict(&x[0]), sparse.predict(&x[0]));

Implementations

impl SparseTsetlinMachine

pub fn from_clauses( clauses: &[Clause], n_features: usize, threshold: f32, ) -> Self

Creates from dense clauses.

Arguments

• clauses - Dense clauses to convert
• n_features - Number of features
• threshold - Classification threshold

pub fn new(clauses: SparseClauseBank, threshold: f32) -> Self

Creates from pre-built sparse clause bank.

pub const fn n_clauses(&self) -> usize

Returns number of clauses.

pub const fn n_features(&self) -> usize

Returns number of features.

pub const fn threshold(&self) -> f32

Returns threshold.

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

Predicts class (0 or 1).

pub fn predict_packed(&self, x: &[u64]) -> u8

Predicts using packed u64 input.

pub fn predict_batch(&self, xs: &[Vec<u8>]) -> Vec<u8>

Batch prediction.

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

Evaluates accuracy on test data.

pub fn memory_stats(&self) -> SparseMemoryStats

Returns memory statistics.

pub fn compression_ratio(&self) -> f32

Returns compression ratio compared to dense model.

Trait Implementations

impl Clone for SparseTsetlinMachine

fn clone(&self) -> SparseTsetlinMachine

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl Debug for SparseTsetlinMachine

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

Formats the value using the given formatter.