Bilevel learning refers to machine learning problems that can be formulated as bilevel optimization models, where decisions are organized in a hierarchical structure. This paradigm has recently gained considerable attention in machine learning, as gradient-based algorithms built on the implicit function reformulation have enabled the computation of large-scale problems involving possibly millions of variables. Despite … Read more
Motivated by the need for reliable traffic management under fluctuating travel demand, we study the problem of determining the worst-case congestion in a multi-commodity traffic network subject to demand uncertainty. To this end, we stress-test a given network by identifying demand realizations and corresponding travelers’ route choices that maximize congestion. The users of the traffic … Read more
We develop a successive, proximal difference-of-convex (DC) function penalty method for solving DC programs with DC constraints. The proposed approach relies on a DC penalty function that measures the violation of constraints and leads to a penalty reformulation sharing the same solution set as the original problem. The resulting penalty problem is a DC program … Read more
Inverse optimization problems are bilevel optimization problems in which the leader modifies the follower’s objective such that a prescribed feasible solution becomes an optimal solution of the follower. They capture hierarchical decision-making problems like parameter estimation tasks or situations where a planner wants to steer an agent’s choice. In this work, we study integral inverse … Read more
ArticleDownload View PDF
We propose a novel approach to modeling fairness in the Vehicle Routing Problem (VRP) by introducing objective functions based on ordering route lengths, capturing both monotonic and non-monotonic equity measures. Our method ensures allocations that are efficient, capacity-feasible, and equitable according to criteria like min-max, range, Gini, variance, or absolute deviations. To prevent biased or … Read more
Decision-dependent robust optimization (DDRO) problems are usually tackled by reformulating them using a strong-duality theorem for the uncertainty set model. If the uncertainty set is, however, described by a mixed-integer linear model, dualization techniques cannot be applied and the literature on solution methods is very scarce. In this paper, we exploit the equivalence of DDRO … Read more
Bilevel optimization is a powerful tool for modeling hierarchical decision-making processes, which arise in various real-world applications. Due to their nested structure, however, bilevel problems are intrinsically hard to solve—even if all variables are continuous and all parameters of the problem are exactly known. Further challenges arise if mixed-integer aspects and problems under uncertainty are … Read more
Transportation network design, or the problem of optimizing infrastructure for a societal goal, subject to individual travelers optimizing their behavior for their own preferences arises frequently in many contexts. However, it is also an NP-hard problem due to the leader-follower or bi-level structure involving a follower objective that is different from yet significantly affects the … Read more
We consider bilevel programs that involve a leader, who first commits to a mixed-integer decision, and a follower, who observes this decision and then responds rationally by solving a linear program (LP). Standard approaches often reformulate these bilevel optimization problems as single-level mixed-integer programs by exploiting the follower’s LP optimality conditions. These reformulations introduce either … Read more