Learning Acceptability Criteria from Good and Bad Decisions: An Inverse Optimization Approach
From MaRDI portal
Abstract: Conventional inverse optimization inputs a solution and finds the parameters of an optimization model that render a given solution optimal. The literature mostly focuses on inferring the objective function in linear problems when acceptable solutions are provided as input. In this paper, we propose an inverse optimization model that inputs several accepted and rejected solutions and recovers the underlying convex optimization model that correctly classifies these given solutions. The novelty of our model is three-fold: First, while most literature focuses on inferring the objective function only, we focus on inferring the feasible region. Second, our model is capable of inferring the constraints of general convex optimization models. Third, the proposed model learns from both accepted (good) and rejected (bad) observations in inferring the constraint set. The resulting model is a mixed-integer nonlinear problem, and to mitigate its complexity, we use the theoretical properties of its solutions to derive a reduced reformulation that is easier to solve. Using a case study on radiotherapy treatment planning for breast cancer patients, we demonstrate that our proposed model can infer a set of clinical guidelines to classify accepted and rejected plans with over 95% accuracy.
This page was built for publication: Learning Acceptability Criteria from Good and Bad Decisions: An Inverse Optimization Approach
Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q6506206)
