Improved excess risk bounds for extreme multi-class supervised contrastive representation learning achieve sample complexity of order R (number of classes) or O(k) (samples per tuple) independent of the rarest class probability.
Furthermore, whenR⩾k, as long asN⩾k×max 164kln 24R ∆ + 100 ln( 12 ∆ ) k(1−τ) , R , we have: sup f∈F | ¯UN(f)−L ϕ(f)|⩽O " CN(H) 1−τ r k N + B √ R (1−τ) 2 r (1− ∥ρ∥ 2
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
citation-role summary
method 1
citation-polarity summary
fields
stat.ML 1years
2026 1verdicts
UNVERDICTED 1roles
method 1polarities
use method 1representative citing papers
citing papers explorer
-
A Refined Generalization Analysis for Extreme Multi-class Supervised Contrastive Representation Learning
Improved excess risk bounds for extreme multi-class supervised contrastive representation learning achieve sample complexity of order R (number of classes) or O(k) (samples per tuple) independent of the rarest class probability.