2 releases

Uses old Rust 2015

0.1.1 Aug 26, 2021
0.1.0 Aug 24, 2021

#1310 in Algorithms

MIT license

9KB
145 lines

label-propagation-rs

Label Propagation Algorithm by Rust.

Label propagation (LP) is graph-based semi-supervised learning (SSL).

A simple LGC and a more advanced CAMLP have been implemented.

Usage

You can find the examples in the examples directory.

The label is a continuous value of [0, class_n], and the result of predict_proba is the value of the label.

use std::result::Result;
use std::error::Error;

extern crate label_propagation;
extern crate ndarray;
extern crate ndarray_stats;

use ndarray::prelude::*;
use label_propagation::{CAMLP, LGC};
use ndarray::Array;

pub fn main() -> Result<(), Box<dyn Error>> {
    // make graph matrix ndarray
    let graph = Array::from_shape_vec(
        (3, 3),
        vec![0.0, 0.3, 0.0, 0.3, 0.0, 0.0, 0.0, 0.0, 0.0]).unwrap();

    // node index for label
    let x = array![0, 1];
    // label
    let y = array![0, 1];

    // make model
    let mut model = CAMLP::new(graph).iter(30).beta(0.1);
    // let mut model = LGC::new(graph).iter(30).alpha(0.99);

    model.fit(&x, &y)?;

    let target = array![0, 1];
    let result = model.predict_proba(&target);
    println!("{:?}", result);

    Ok(())
}

develop

docker build -t graph .
docker run -it -v $PWD:/app graph bash

Thanks

Dependencies

~3.5MB
~59K SLoC