Presentation on YouTube.

Combinatorial optimisation is used to solve essential problems in industry and society. It excels on well-defined problems, where highly efficient constraint solvers and algorithms can be used to find optimal or near-optimal solutions. But what if the problem is less well-defined? From an AI perspective, we hope to build intelligent systems that can handle such situations by levering relevant data or by directly interacting with a user. I will provide examples and recent developments in decision-focussed learning where regression/forecasting models are built by backpropagating over (non-differentiable) solvers; in preference learning/inverse optimisation over historic solutions and preference elicitation from users; as well explainable constraint solving and Large Language Model based modeling support tools. The research is part of my ambitious 5 year ERC Consolidator grant ‘Conversational Human-Aware Technology for Optimisation’ (CHAT-Opt).

Bio: Tias Guns is Associate Professor in Computer Science at the DTAI lab of KU Leuven, Belgium. Tias' expertise is in the hybridisation of machine learning systems with constraint solving systems, more specifically building constraint solving systems that reason both on explicit knowledge as well as knowledge learned from data. He aims to make constraint solving technology more accessible, as well as to make the technology more human-aware by learning from the daily operational environment and its users. He was awarded a prestigious ERC Consolidator grant in 2021 to work on conversational human-aware technology for optimisation and currently leads a lab of 8 PhD students and 4 postdocs.