CR-BLEA uses an online-trained contrastive ranking network to allocate lower-level optimization resources only to promising upper-level solutions, reducing function evaluations in bilevel evolutionary algorithms by about a third while preserving accuracy.
Camacho-Vallejo, C
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.NE 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
CR-BLEA: Contrastive Ranking for Adaptive Resource Allocation in Bilevel Evolutionary Algorithms
CR-BLEA uses an online-trained contrastive ranking network to allocate lower-level optimization resources only to promising upper-level solutions, reducing function evaluations in bilevel evolutionary algorithms by about a third while preserving accuracy.