Hélène Verhaeghe (UCLouvain)

Title : The best of both worlds: when constraint programming and machine learning help each other

Abstract : Constraint programming is well known to be performant at finding the solutions of combinatorial optimization problems and can provide guarantees about these solutions. However, as problems tend to grow in size and complexity, the limits of such methods tend to show. Machine learning, on the other hand, is very good at extracting statistical patterns and dealing with uncertainty. However, they are really bad at reasoning, even on simple tasks. This talk is about how, by combining the two families of techniques, they can benefit one another. This talk will start with an introduction to constraint programming for those not knowledgeable about the technique.

The seminar will take place in Room S08 at the Faculty of Sciences.