A robust approach for modeling limited observability in bilevel optimization

Many applications of bilevel optimization contain a leader facing a follower whose reaction deviates from the one expected by the leader due to some kind of bounded rationality. We consider bilinear bilevel problems with follower's response uncertainty due to limited observability regarding the leader's decision and exploit robust optimization to model the decision making of the follower. We show that the robust counterpart of the lower level allows to tackle the problem via the lower level's KKT conditions. (C) 2021 Elsevier B.V. All rights reserved.

Automated summary

In applications of bilevel optimization, the follower's response uncertainty is crucial due to deviations from the leader's expectations. By utilizing robust optimization to model the follower's decision-making, we can solve the problem using the KKT conditions of the lower level.