Struct SampleComplexity 

pub struct SampleComplexity {
    pub eps: f64,
    pub delta: f64,
    pub vc_dim: usize,
}

Sample complexity for PAC learning (Blumer et al. / Vapnik-Chervonenkis bound).

Returns m = ceil((d * ln(d/eps + 1) + ln(2/delta)) / eps) samples. Here vc_dim is the VC dimension d of the hypothesis class.

Fields

eps: f64

Accuracy parameter ε ∈ (0,1).

delta: f64

Confidence parameter δ ∈ (0,1).

vc_dim: usize

VC dimension d of the hypothesis class.

Implementations

impl SampleComplexity

pub fn new(eps: f64, delta: f64, vc_dim: usize) -> Self

Create a new SampleComplexity instance.

pub fn compute(&self) -> usize

Compute the sample complexity upper bound.