Time filter

Source Type

Toulouse, France

Bazhar Y.,University of Poitiers | Ait-Ameur Y.,IRIT ENSEEIHT | Jean S.,University of Poitiers | Baron M.,University of Poitiers
WEBIST 2012 - Proceedings of the 8th International Conference on Web Information Systems and Technologies | Year: 2012

Several ontologies are currently defined for various domains and for a wide range of applications. As a consequence, two problems shall be addressed (1) the size of data described by ontologies is huge and (2) existing ontologies are not all similar: they can be defined with different formalisms leading to different types of ontologies (canonical, non canonical and linguistic ontologies). The first problem has been solved by using databases to store data and ontologies that define the semantics of data. However, these databases, called Ontology-Based Databases (OBDB), do not handle the second problem i.e., that different types of ontologies exist. Indeed, they usually support a given ontology formalism without providing any mean to handle ontologies defined in an other formalism. In this paper, we propose an extension of the OntoDB OBDB to support the introduction of new operators that could be exploited by its exploitation language OntoQL. Such operators can be implemented by an external program located outside the database, or could invoke a web service. We show that these operators can be used to introduce new ontology constructors in order to support a given ontology model inside OntoDB. As a running example, we consider the support of the OWL non canonical constructors which are not supported in the native OntoDB OBDB. Source

Meyer N.,University Paul Sabatier | Lauwers-Cances V.,University Paul Sabatier | Lauwers-Cances V.,Toulouse University Hospital Center | Lourari S.,University Paul Sabatier | And 8 more authors.
British Journal of Dermatology | Year: 2014

Background Early diagnosis and rapid surgical excision are essential for improving the prognosis of patients with melanoma. Reflectance confocal microscopy has been validated as a feasible procedure for in vivo diagnosis of melanoma but cannot be used to measure tumour thickness. However, ultrasonography and optical coherence tomography may allow melanoma thickness to be measured in vivo. Objectives To validate the accuracy and reliability of high-frequency ultrasonography (HFUS) and optical coherence tomography for assessing melanoma thickness in vivo. Methods We conducted a prospective study on 131 patients with at least one equivocal melanocytic lesion. Each lesion underwent optical coherence tomography and HFUS assessment, followed by excision and pathological examination. Histopathology was considered to be the gold standard for assessing melanoma thickness. Repeatability, inter- and intrarater reproducibility and reliability were evaluated for each imaging procedure. Results Ultrasonography showed a good level of agreement with histology [intraclass correlation coefficient (ICC) 0·807; 95% confidence interval (CI) 0·703-0·877] and excellent inter-rater reproducibility (G = 0·97), resulting in reliable in vivo assessment of melanoma thickness. The 930-nm optical coherence tomography showed a poor level of agreement with histopathology (ICC 0·0; 95% CI -0·2-0·2) and the inter-rater reproducibility was null (G = 0·00). Conclusions HFUS is a reliable and reproducible noninvasive method for assessing melanoma thickness. Routine use of HFUS may allow single-step excision of equivocal melanocytic lesions, with surgical margins determined by in vivo assessment of tumour thickness. What's already known about this topic? Early diagnosis and complete surgical excision with defined margins, depending on tumour thickness, is the recommended strategy for improving melanoma prognosis. Most primary melanomas require a two-step surgical procedure, with excision of the tumour followed by re-excision determined by histopathological assessment of tumour thickness. What does this study add? High-frequency ultrasonography is a reliable tool that provides reproducible results for in vivo evaluation of melanoma tumour thickness. This method could enable single-step surgical excision of melanoma. © 2014 British Association of Dermatologists. Source

Ayed R.,Manouba University | Bounhas I.,University of Tunis | Elayeb B.,Manouba University | Evrard F.,IRIT ENSEEIHT | Saoud N.B.B.,Manouba University
Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) | Year: 2012

This paper proposes and experiments a new approach for morphological feature disambiguation of non-vocalized Arabic texts using a possibilistic classifier. The main idea is to learn contextual dependencies between features from vocalized texts and exploit this knowledge to disambiguate non-vocalized ones. We use possibility theory as a means to model imprecision in the training and testing steps, since the context is itself ambiguous. We also investigate the dependency between various features focusing on the Part-Of-Speech (POS). © 2012 Springer-Verlag. Source

Zaibi G.,University of Sfax | Peyrard F.,IRIT ENSEEIHT | Kachouri A.,University of Sfax | Fournier-Prunaret D.,INSA Toulouse | Samet M.,University of Sfax
Security and Communication Networks | Year: 2014

Information security using chaotic dynamics is a novel topic in the wireless sensor network (WSN) research field. After surveying analog and digital chaotic security systems, we give a state of the art of chaotic S-Box design. The substitution tables are nonlinear maps that strengthen and enhance block crypto-systems. This paper deals with the design of new dynamic chaotic S-Boxes suitable for implementation on wireless sensor nodes. Our proposed schemes are classified into two categories: S-Box based on discrete chaotic map with floating point arithmetic (cascading piecewise linear chaotic map and a three-dimensional map) and S-Box based on discrete chaotic map with fixed-point arithmetic (using discretized Lorenz map and logistic-tent map). The security analysis and implementation process on WSN are discussed. The proposed methods satisfy Good S-Box design criteria and exceed the performance of Advanced Encryption Standard static S-Box in some cases. The energy consumption of different proposals and existing chaotic S-Box designs are investigated via a platform simulator and a real WSN testbed equipped with TI MSP430f1611 micro-controller. The simulations and the experimental results show that our proposed S-Box design with fixed-point arithmetic Lorenz map has the lowest energy-consuming profile compared with the other studied and proposed S-Box design. © 2013 John Wiley & Sons, Ltd. Source

Kim J.-S.,Carnegie Mellon University | Gurdjos P.,IRIT ENSEEIHT | Kweon I.S.,KAIST
Computer Vision and Image Understanding | Year: 2010

We propose confocal conics as a new planar pattern for camera calibration with poses. We show that confocal conics encode the metric information which can be easily extracted with their generalized eigenvalues. The proposed method is tested with synthetic and real data, and shows comparable performances with conventional algorithms, while it still has all virtues of curved features. © 2010 Elsevier Inc. Source

Discover hidden collaborations