Maître de conférences
Université de Lille
Bâtiment P2 - Bureau 205
Homepage : https://john-klein.github.io
Adresse: Cité Scientifique
Avenue Carl GAUSS
59650 Villeneuve d’ascq
Téléphone: (+33) 03 20 43 48 77
Research Interests :
Ill known probabilities with applications to information fusion, signal processing or machine learning.
Real life data are tainted with many imperfections (uncertainty, imprecision, ambiguity, incompleteness, inconsistency and so on..). What can data scientists do with pieces of information like
- This guy is about 30 years old
- The target is in volume V with a probability greater than 0.8
In both examples, one wishes to estimate an unknown quantity (age of the guy, 3D position of the target). There is no obvious way to characterize these quantities by a single probability measure.
Instead, one may relax single probability representations and characterize the unknown quantities by sets of probabilities or by probabilities of probabilities. This is what ill-known probabilities are about.
Related frameworks are for instance
- imprecise probabilities,
- belief functions,
- random measures,
- robust Bayes.
I have also interests in machine learning and especially ensemble methods which are also falling in the scope of the information fusion paradigm. Ensemble methods, or multiple classifier systems are a class of machine learning technique in which training algorithms are run several times. Most of the time, the training algorithms are meant to converge to different and complementary decision functions. This can be achieved by training on different data points, features or with different algorithms.
Teaching activities :
I am in charge of the following courses at the department of electronics, electrical engineering and automation of Lille1
- Signal Processing (Master ASE - 1st year)
- Mobile Robotics project - initiation (Master ASE - 1st year)
- UML (Master GI - 1st year)
- Real Time Systems (Master GI - M2 II)
- Software Engineering (Master GI - 1st year)
Since 2016, I am in charge of the following course at the department of computer sciences of Lille1
- Machine Learning (Master Informatique - M2 MOCAD)
Finally, I am also in charge of the following course at Ecole Centrale de Lille
- Data Fusion (Final year students)