Struct Perceptron 

pub struct Perceptron {
    pub weights: Vec<f64>,
    pub bias: f64,
    pub mistakes: usize,
}

Online Perceptron classifier.

Fields

weights: Vec<f64>

Weight vector w ∈ ℝ^d.

bias: f64

Bias term b.

mistakes: usize

Number of mistakes made so far.

Implementations

impl Perceptron

pub fn new(dim: usize) -> Self

Create a new zero-initialized Perceptron.

pub fn predict(&self, x: &[f64]) -> f64

Predict the label for input x: sign(w·x + b).

pub fn update(&mut self, x: &[f64], label: f64) -> bool

Online update: if prediction wrong, w ← w + y·x, b ← b + y.

pub fn mistake_bound(radius: f64, margin: f64) -> usize

Perceptron mistake bound: M ≤ (R/γ)² where R = radius, γ = margin.