Robust Optimisation

Robust optimization focuses on finding solutions to optimization problems that remain effective even when faced with uncertainties in input parameters or model assumptions. Current research emphasizes developing algorithms that handle multi-objective problems and those with unknown model hyperparameters, often employing Bayesian optimization or novel approaches like double implicit layer models for robust convex QCQPs. These advancements are crucial for improving the reliability and trustworthiness of machine learning models in various applications, particularly in safety-critical domains where robustness is paramount, and for enhancing the accuracy and efficiency of likelihood-free inference methods. The development of provably robust and plausible counterfactual explanations further contributes to the explainability and trustworthiness of these models.