Table 1 Notation and Symbols

r ∈ R

Dataset of image patches

t ∈ {0, ⋯, T}

Iteration of ActiveLearn

S^tr, S^te

Unlabeled training, testing pools

Φ

Training methodology

$S_t^{\mathsf{\text{E}}},{\widehat{S}}_t^{\mathsf{\text{E}}}$

Eligible samples, annotated samples

$S_{t,\Phi }^{\mathsf{\text{tr}}}$

Samples labeled via Φ at t

${\mathcal{T}}_t$

Fuzzy classifier using $S_{t,\Phi }^{\mathsf{\text{tr}}}$

k_1,t, k_2,t

Number of samples in $S_t^E$ from ω₁, ω₂

M

Number of votes used to generate ${\mathcal{T}}_t$

ω₁, ω₂

Possible classes of r

τ

Confidence margin

r ↪ ω ₁

Membership of r in class ω₁

θ

Classifier-dependent threshold for ${\mathcal{T}}_t$

$\widehat{k_1},\widehat{k_2}$

Number of samples in ${\widehat{S}}_t^{\mathsf{\text{E}}}$ from ω₁, ω₂

p _t (r ↪ ω₁)

Probability of observing r ↪ ω₁

N _t

Samples added to training set at t

P _Δ

Model confidence

${\widehat{P}}_t$

Probability of observing $\widehat{k_1}$ samples

${\mathcal{A}}_t$

Accuracy of trained classifier at t

$ℒ$

Total training cost after T iterations