| |
|
Table of contents
Volume
9
Number 1
|
Editorial
Mohand Boughanem, Salem Benferhat, Guy Melançon
Causal Discovery in Non-Ideal Framework
Stijn Meganck, Philippe Leray, Bernard Manderick
Détection d'attaques élémentaires et coordonnées à base de réseaux Bayésiens naïfs
Tayeb Kenaza, Karim Tabia, Aicha Mokhtari
Une inférence lexicographique à partir de bases de croyances partiellement préordonnées
Safa Yahi, Sylvain Lagrue, Mariette Sérayet, Odile Papini
All abstracts are in PDF format.
|
|
|
Editorial
We are proud to publish this issue of the Information - Interaction - Intelligence (I3) international journal. The present issue
contains three papers: two of them are written in French, while the third one is written in English .
The first paper, in English, and entitled
"Causal Discovery in Non-Ideal Frameworks" is by Stijn Meganck, Philippe Leray and Bernard Manderick. This paper deals with an important problem in Artificial Intelligence area and concerns the handling of causal
information. The authors are interested in using graphical models frameworks. Recently, the concept of intervention or manipulation, symbolised by the "do" operator, has been introduced in order to
distinguish causality from a mere correlation. The authors propose two approaches for discovering causal relations between variables, using
experimental and interventional data (and not only passive data obtained from observations). These two approaches are particularly
appropriate when insufficient data is present to perform statistical tests reliably and when there are latent variables respectively.
The second paper, in French, also concerns graphical models and bayesian networks. It is entitled “Détection d'attaques
élémentaires et coordonnées à base de réseaux Bayésiens naïfs", by Tayeb Kenaza, Karim Tabia and Aicha Mokhtari. Bayesian networks
are important and powerful tools to reason under uncertainty. The authors study two approaches, based on naive bayes, applied to the problem of detecting elementary and coordinated attacks. Some of the
results contained in this paper won "best student paper award" in SECRYPT 2008 conference (International Conference on Security
and Cryptography), held in Porto, July 2008. In this paper, the two proposed approaches take advantage of available data to provide
efficient algorithms for predicting plausible attacks. The authors also show how decision rules, used in naive bayes classifiers, can be
improved to detect novel attacks and novel normal activities.Experimental results, done on both existing benchmarks and real data,
show the efficiency of the proposed approaches.
The third paper, in French, entitled "Une inférence
lexicographique à partir de bases de croyances partiellement préordonnées"
by Safa Yahi, Sylvain Lagrue, Mariette Sérayet and Odile Papini. This paper, which is an extended version of a communication
presented at IAF-08, deals with the problem of handling inconsistencies in belief bases. The authors proposed two solutions to
analyse the lexicographic inference, or cardinality-based inference, in
situations where available pieces of information are only partially preordered.
The first solution is based on the concept of totally preordered
compatible bases while the second solution directly compares
coherent sub-bases of the initial inconsistent base. The two solutions
are illustrated with an example inspired from an application within the
framework of the European VENUS project regarding managing
archaeological information.
Mohand Boughanem, Salem Benferhat et Guy Melançon
|
|