Every day, many problems of resource allocation, planning and scheduling, also known as constraint satisfaction or optimization problems, are solved by advanced software that does not exploit the vast amount of data that are continuously generated by the technology supporting the processes and activities that are subject of optimization. One remarkable example is society, where human activities mediated by the ICT's leave huge amounts of digital traces: such big data could act as feedback into constraint models, to devise better solutions that can adapt to actual people behavior - e.g., a public transportation schedule that continuously adapts to the real mobility patterns of people represented by the digital traces of travels. Therefore, the knowledge represented in (big) data could help the adaptation of schedules, resources and plans coherently with the real dynamics in the real world.
So far, constraint solving has evolved quite independently from machine learning and data mining. Very recently, interest has been growing on the integration of these two fields, which can work in two ways: (a) constraint solvers can be included in machine learning and data mining algorithms; and (b) machine learning and data mining can help in addressing constraint problems.
Promising initial results have been achieved in both directions, but most aspects require further research for establishing a fruitful integration.
The summer school "Constraint Programming meets Data Mining" provides an intensive training opportunity to learn the essentials of recent research on constraint solving, machine learning and data mining, and the key aspects related to their integration.
Student will follow lectures from top experts of the fields, and will receive personalized training on selected exercises in hands-on labs.