Bertet Karell > Karell Bertet - eng
email : Karell.bertet univ-lr.fr
Associtate professor (habilitation)
Research team : IDDC
Tel : 05.46.45.83.08
Website (in progress) (http://perso.univ-lr.fr/kbertet/)
Research interest:
Formal Concept Analysis and images:
Since the 2000s, the emergence of formal concept analysis (FCA) in various areas of computing, as in data mining and knowledge representation for databases and ontologies, has highlighted the structures of concept lattice and closed set lattice. The emergence of the lattice structure can be explained both by the growing share of computer science in most disciplines, leading to increasing production data, but also by the increasing power of computers.
The concept lattice is a graph having the property of lattice, and defined from data organized as a binary table object x attribute.
The vertices are concepts - a concept is a maximum group of objects with common attributes - and the lattice is composed of concepts in relation by inclusion.
My work in this context consists in using developping tools of FCA for images data.
Structural and algorithmics aspects of lattices:
The first reference book on lattice theory is due to Birkhoff in 1940 (first edition). However, the structure of lattice has been introduced in the late 19th century as an algebraic structure equipped with two operators called lower and upper bound.
In an application framework, efficient operation of the existing algorithms to manipulate these structures requires a good knowledge of the formalism and structural properties of the lattice theory to identify algorithmic tools suitable for a given problem.
My work in this context concerns the structural and algorithmic study of the canonical direct basis, a rule system with interesting properties.
Results :
Formal Concept Analysis and images:
- Navigala, a method of classification by navigation in a concept lattice
- Structural signature for images of symbols
- Study of dichotomic lattices
Structural and algorithmics aspects of lattices:
- Study od the canonical direct basis
- Lattice library in Java
Main publications (4 last years) :
Revues:
- M. Coustaty, K. Bertet, M. Visani, et J.-M. Ogier. A New Adaptive Structural Signature for Symbol Recognition by Using a Galois Lattice as a Classifier. IEEE Transactions on Systems, Man, and Cybernetics (TSMC), May 2011
- M. Visani, K. Bertet and J-M. Ogier. Navigala. An Original Symbol Classifier Based on Navigation through a Galois Lattice. International Journal on Pattern Recognition and Artificial Intelligence (IJPRAI), 2011.\
- K. Bertet and B. Monjardet. The multiple facets of the canonical direct basis. Theoretical Computer Science. 411(22-24):2155-2166, ISNN 0304-3975, May 2010
- K. Bertet, M. Visani, N. Girard. Treillis dichotomiques et arbres de décision. Traitement du Signal, numéro spécial, volume 26, numéro 5, p 407-416, 2009.
Lecture notes:
- M. Rusinol, K. Bertet, J.-M. Ogier et J. Llados. Symbol recognition by using a concept lattice of graphical patterns. Graphics Recognition: Recent Advances and New Opportunities, Lecture Notes in Computer Science (LNCS), (special issue post GREC’09). Springer-Verlag ed., Vol 6020, 2010, p.187-198. \
- M. Coustaty, S. Guillas, M. Visani, K. Bertet et J.-M. Ogier. On the joint use of a Structural Signature and a Galois Lattice Classifier for Symbol Recognition. Graphics Recognition: Recent Advances and New Opportunities, Lecture Notes in Computer Science (LNCS), (special issue post GREC’07). Springer-Verlag ed., Vol. 5046, 2008, p. 61-70.
- S.Guillas, K. Bertet, et J.-M. Ogier. Towards an iterative classification based on concept lattice. Concept Lattices and their Application, Lecture Notes in Artificial Intelligence (LLNAI), (special issue post CLA’06). Springer-Verlag ed., Vol. 4923, p 256-262, Feb 2008.
Teaching (since 2008):
- Master Icone: Graph theory and complexity ; Association rules ; Formal Concept Analysis; Ontologies
- Licence: Experts systems ; Syntaxic analysis ; Méthodologie ; Databases