AI-Fun with ELLIS Seminar | Theo Damoulas: Decision Making under the Exponential Family: Distributionally Robust Optimisation with Bayesian Ambiguity Sets
| Dates: | 18 June 2025 |
|---|
| Times: | 11:00 - 12:00 |
|---|
| What is it: | Seminar |
|---|
| Organiser: | Faculty of Science and Engineering |
|---|
| Who is it for: | University staff, External researchers, Alumni, Current University students |
|---|
| Speaker: | Theo Damoulas |
|---|
|
The Manchester Centre for AI Fundamentals and Manchester's ELLIS Unit are co-hosting a series of seminars featuring expert researchers working in the fundamentals of AI.
On 18 June, we are joined by Prof Theo Damoulas from the University of Warwick.
Title: Decision Making under the Exponential Family: Distributionally Robust Optimisation with Bayesian Ambiguity Sets
Abstract: Decision making under uncertainty is challenging as the data-generating process (DGP) is often unknown. Bayesian inference proceeds by estimating the DGP through posterior beliefs on the model's parameters. However, minimising the expected risk under these beliefs can lead to suboptimal decisions due to model uncertainty or limited, noisy observations. To address this, we introduce Distributionally Robust Optimisation with Bayesian Ambiguity Sets (DRO-BAS) which hedges against model uncertainty by optimising the worst-case risk over a posterior-informed ambiguity set. We provide two such sets, based on posterior expectations (DRO-BAS(PE)) or posterior predictives (DRO-BAS(PP)) and prove that both admit, under conditions, strong dual formulations leading to efficient single-stage stochastic programs which are solved with a sample average approximation. For DRO-BAS(PE) this covers all conjugate exponential family members while for DRO-BAS(PP) this is shown under conditions on the predictive's moment generating function.
Bio: I am a Professor in Machine Learning and Data Science with a joint appointment in Computer Science and Statistics. I am a Turing AI Fellow (2021-2026) having received the UK Research and Innovation (UKRI) Turing AI Acceleration Fellowship in order to lead research on setting the Machine Learning Foundations of Digital Twins. I am an ELLIS member and also affiliated with New York University as a Visiting Exchange Professor at the Center for Urban Science and Progress (CUSP). My research interests are in probabilistic machine learning and Bayesian statistics with an emphasis on the study and integration of various forms of structure and inductive biases (structured priors, spatiotemporal dependencies, dynamics, compositions, physical laws, flows, causality, etc) while advancing robust and scalable approximate inference methodologies. My research has broad applications in Digital Twins, Bayesian nonparametrics and spatiotemporal problems in urban science and computational sustainability. I am the founder and PI of the cross-departmental Warwick Machine Learning Group and I lead two large projects at The Turing that are impact stories.
This seminar is free to attend: you can come in person to LT1.4, Kilburn Building or join online. To get the zoom link, please register at the Ticketsource link on this page.
Travel and Contact Information
Find event
Lecture Theatre 1.4
Kilburn Building
Manchester
