Workshop on Bayesian Optimisation with Multiple Objectives: Open Challenges for Machine Learning and Optimisation
Monday 27 February, 10am-6pm, Scarman Conference Centre, University of Warwick
In collaboration with the Alan Turing Institute, Warwick is hosting a one-day workshop for researchers on multi-objective Bayesian optimisation to discuss latest advances in the field, explore challenges and build a community network.
The workshop is aimed at researchers in multi-objective Bayesian optimisation, but also welcomes academics from other fields wishing to optimise expensive black-box multi-objective problems, as well as machine learning (ML) researchers interested in Bayesian surrogate models and combining ML and optimisation.
Bayesian optimisation is one of the great successes of ML. It's widely applied in industry as it enables the optimisation of problems that would be otherwise intractable. It uses machine learning models to approximate the objective function of an expensive black-box optimisation problem, then decides which additional information would be most valuable.
Typical application examples include engineering design (where the evaluation of a candidate solution requires an expensive simulation, perhaps from a digital twin of the real system), or carrying out a physical experiment, and hyperparameter tuning (where the evaluation of a hyperparameter setting requires the training of an ML model).
Many real-world problems require the simultaneous optimisation of conflicting objectives. For example, in ML models, there may be a trade-off between accuracy and fairness, in engineering there may be the trade-off between performance and design robustness, meaning that extending Bayesian optimisation to multi-objective problems is of great practical interest.
Professor Ruth Miseneris a Professor in Computational Optimisation in Imperial College London's Department of Computing. The foundations of Ruth's research lie in numerical optimisation algorithms and computational software. Her applications focus on optimization challenges arising in industry, e.g. scheduling in manufacturing or experimental design in chemicals research. Ruth also contributes at the interface between operations research and machine learning. Ruth received an SB from MIT (2007) and her PhD (2012), from Princeton.
Ruth is the BASF/RAEng Research Chair in Data-Driven Optimisation (2022-27). She received the Macfarlane Medal as the overall winner of the 2017 RAEng Engineers Trust Young Engineer of the Year competition.
Her work has been recognized with best paper awards from: the Journal of Global Optimization (2013), International Conference on Autonomous Agents & Multi-Agent Systems (Best Innovative Demo, 2020), Conference on the Integration of Constraint Programming, Artificial Intelligence, & Operations Research (2021), and Optimisation & Engineering (2021). Ruth’s research team develops popular open-source code. The Optimisation & Machine Learning Toolkit (OMLT) she worked on won the 2022 COIN-OR Cup for its contribution to open-source operations research software development.
Sam Daulton, University of Oxford
Sam Daulton is a research scientist at Meta, a DPhil candidate in machine learning at the University of Oxford, and co-creator of BoTorch—an open-source library for Bayesian optimisation research.
Sam works with Eytan Bakshy and Max Balandat at Meta and Mike Osborne at Oxford. His current research focuses on methods for Bayesian optimisation in challenging scenarios. Previously, Sam worked with Finale Doshi-Velez at Harvard University on efficient and robust methods for transfer learning.