Today, based on the advances in IT and digital data storage, in many industrial, economic, medical or other application areas, increasing amounts of signals, measurements, images and other types of data become available, implicitly describing underlying processes or structures. With this availability the potential - and need - arises for advanced intelligent tools to extract the underlying information, predict, diagnose, estimate or make use of it in some other way, in order to optimize or improve services. Since structure in this data is mostly hidden under noise, due to the stochastic nature of the processes and their measurement, robust and adaptive tools are needed that can cope with this nature.
"Computational intelligence and learning" intends answering to this need; it gathers research work carried out in various disciplines, with the objective of adding some form of intelligence or automatic learning of situations and properties in algorithms, data processing tasks, data mining, information extraction, etc. Computational intelligence and learning concerns disciplines and concepts such as machine learning, artificial neural networks, data mining, fuzzy logic, evolutionary computation, probabilistic techniques. It is a specific scientific domain on its own, as attested by the existence of several international scientific societies covering part or all of the topics: the IEEE (Institute of Electrical and Electronics Engineers) Computational Intelligence Society, the International Neural Network Society, the European Neural Network Society, the ACM SIGKDD Special Interest Group in Knowledge Discovery and Data Ming, the International Machine Learning Society, and many others.
A marked characteristic of this field is its multidisciplinarity. It has been motivated in part by biologists and was developed by researchers and practitioners in computer science, statistics, engineering and others. This has resulted in a variety of successful computational tools such as artificial neural networks, fuzzy logic, evolutionary computation, swarm intelligence, classifier induction, reinforcement learning, etc. Many of these methods are used today in a wide variety of application domains, from engineering to finance, through data analysis, chemometrics, signal processing, economy and many others. Indeed, many recent developments in machine learning would not have been possible without biological inspiration, statistical background, specific problems brought by engineers, and solutions implemented by computer scientists. The multidisciplinary and evolving character of the field is without any doubt the main ingredient of its richness.
Allowing the PhD students to benefit from this multidisciplinarity is the goal of this graduate school. On the "Communaute Francaise de Belgique" side, the graduate school will be based on the recently created "groupe de contact FNRS" in Machine Learning; on the "Vlaamse Gemeenschap" side, several promoters of this graduate school are members of the FWO-sponsored "Research Network Machine Learning for Data Mining and its Applications". The graduate school extends the scope of these networks though, in addition to a larger geographic coverage (it will cover Belgium, and have contacts with foreign institutions).
Belgian researchers in Machine Learning and Computational Intelligence are widely renowned. Some topics are studied since 20 years, making the researchers pioneers in their respective areas. Moreover, there is currently a large, and increasing, number of PhD researchers directly concerned by the topics of this graduate school. The coordinators of this initiative are convinced of its cross-fertilization added value. It is expected that bringing together PhD students having common concerns and interests, but working in different fields such as statistics, data mining, computer science, applied mathematics, systems and control, signal processing and other areas, will enhance their research abilities and knowledge.
Notice however that several teams at the basis of this graduate school initiative are also involved in other graduate school: there are other graduate schools in statistics, computer science, systems and control, signal and communications, and others. PhD students in these teams will be able to benefit from the activities of these graduate schools too. Nevertheless, without the graduate school in Computational Intelligence and Learning, the essential multidisciplinary character would be largely missed. It is therefore not in the intentions of the graduate school in Computational Intelligence and Learning to replace or prevent any other initiative. On the contrary, in addition to own initiatives, the Computational Intelligence and Learning graduate school is a complementary, transversal project that will strengthen and open to a wider public, the initiatives taken by other graduate schools.