Predict and Optimize: combinatorial optimisation with machine-learning based input
Combinatorial optimisation is used to solve large scale routing, planning and scheduling problems. But how do these techniques work? And what if not all data is known, and want to integrate machine learning predictions?
Practical information:
Want to register?
- Register until: 16 Mar 2023
- Price: Free
Do you need a certificate of attendance to this course? Send us an email and we will take care of it.
Participation is free but registration is mandatory. Please register in order to receive the link to the course. This will be sent the day of the seminar 👇
Georganiseerd door:
Leertraject
In this course, we review different combinatorial optimisation problems (Vehicle Routing, Scheduling and combinatorial puzzles) as well as discrete optimisation techniques for solving them, including Branch and Bound, Integer Programming, Dynamic Programming and Multi-Valued Decision Diagrams, Constraint Programming, Local Search and Hybridization of the aforementioned techniques. A focus will be on the reusability and genericity of the techniques for solving large classes of constrained optimisation problems.
With a better understanding of the solving techniques, we then move
to the second part of the course which focuses on ‘Predict and Optimize’
problems, where part of the input to the combinatorial optimisation
problem has to be inferred from data. We will review issues such as
perception-based solving (e.g. visual sudoku), learning preferences in
vehicle routing and end-to-end learning for energy-aware scheduling. We
will discuss how to improve either the learning or the solving, as well
as the integration of the two to obtain the best results.
Part 1 will be taught by Pierre Schaus, UC Louvain, using material borrowed from his new MOOC on constraint programming.
Part 2 will be taught by Tias Guns, KU Leuven. Links will be provided to example notebooks in the CPMpy constraint modeling library (Github).
Teachers / speakers
Pierre Schaus
Pierre Schaus is a professor of computer science at UCLouvain since 2012 where he teaches the constraint programming course since 2017. He has worked on several constraint programming solvers, used CP to solve real-life applications, and published scientific papers in the major conferences of the CP field (CP, CPAIOR, etc).
Tias Guns
Tias Guns is an AI researcher working at the intersection of data science and constraint optimisation. Constraint solving is a key technology for solving scheduling, planning and configuration problems across all industries.
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. For example learning the preferences of planners in vehicle routing, and solving new routing problems taking both operational constraints and learned human preferences into account; or building energy price predictors specifically for energy-aware scheduling, and planning maintenance crews based on expected failures.
This has the potential to capture both the problem structure and more subjective aspects such as human preferences and changing environments. The ultimate goal is to make constraint solving techniques more intelligent and human-aware.
VAIA and Trail Joint Seminar Series for researchers
In this seminar series, we bring together researchers that are interested in, or conducting research on AI and Machine Learning. Each VAIA-TRAIL doctoral course focuses on a specific topic ranging from times series to reinforcement learning to combat epidemics and interpretable & explainable Deep Learning.
Related courses
Summer School on Security and Privacy in the Age of AI
Summerschool - Heverlee - KU Leuven, UGent, VUB, Imec