Clarifies relationships among MPEC constraint qualifications and fully characterizes MPEC-LICQ for the MPEC from bilevel hyperparameter optimization in L1-loss SVM classification.
Meanwhile, ( BB ⊤)(Λc 3,Λu)1|Λu| is replaced by 28 (BB ⊤)(Λc 3,Λc 3∪Λu)1|Λc 3∪Λu|
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
math.OC 1years
2025 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
On Constraint Qualifications for MPECs with Applications to Bilevel Hyperparameter Optimization for Machine Learning
Clarifies relationships among MPEC constraint qualifications and fully characterizes MPEC-LICQ for the MPEC from bilevel hyperparameter optimization in L1-loss SVM classification.