Topic: versatile models for relational data and unsupervised learning tasks. Supervisor: Dr. Nicolas Lachiche.
Keywords: Data mining, spatiotemporal data mining, Spatial analysis, Geographical Information systems, Machine learning, Predictive systems, Land cover/use maps, remote sensing.
Title: Data Mining Approach for Spatiotemporal Relationships Extraction: Application to Natural Risks prediction.
Thesis Supervisor: Prof. Sami Faiz (Full Professor, ISAMM, Tunisia)
Summary: Various studies have been directed to capture time-varying characteristics of spatial data using data mining techniques. Nevertheless, the exploration of these data sets with classical and statistical techniques is a challenging task especially with the existence of many interesting hidden relations and patterns. The present thesis proposes a new approach for the prediction of future spatiotemporal relationships based on spatiotemporal association rules. Such rules demonstrate the evolution of spatial objects and the influence of the spatial distribution of adjacent-areas. Our approach is based on two phases. The first one aims to extend association rules mining techniques to spatiotemporal data. We are interested in looking for spatiotemporal association rules that relate properties of reference objects like large towns with properties of other spatial task-relevant objects. Each relevant object is typically a spatial layer. The extracted patterns are relationships relating the spatial objects (reference objects and relevant objects) during different time periods. This first phase is composed of four steps; the core step is to extract the spatiotemporal relationships, it is handled by spatial queries taking into account time period indication. The second and third steps are devoted respectively to frequent spatiotemporal itemsets generation and spatiotemporal association rules extraction. The fourth step is the refinement of the extracted rules. We propose two measures to evaluate the quality of the obtained spatiotemporal association rules. On the basis of the proposed measures namely the spatial closeness relevance and the time subsequence, only interesting spatiotemporal association rules are retained. The second phase aims to generate spatiotemporal predictive rules. We use a predictive neural network based on a nonlinear time series technique to generate these rules. The learning examples correspond to spatiotemporal association rules and the results are spatiotemporal predictive rules assessing the future spatiotemporal relationships. Such relationships are not to be neglected; they can be a major issue influencing natural risks occurrences. We conduct an experimentation using time series of satellite images describing Megrine zone in the southern coast of Tunis (Tunisia). As a final result, we obtain spatiotemporal predictive rules describing the spatiotemporal relationships evolution between anterior and future dates. The evaluation results show good performance of the proposed approach measured in terms of precision, accuracy, sensitivity, sensibility and other interest measures. Keywords: Spatiotemporal data mining, spatiotemporal association rules, spatiotemporal relationships, neural network, time series, natural risks prediction.
Below the main teached themes :
* Optimization of merchant web sites
Audience: 2nd year students of Professional Master in Electronic Commerce at FSJEGJ (Lecture) Period: 2012-2013, 2013-2014, 2014-2015 and 2015-2016.
* Methodologies of SI conception: unified processes
Audience: 1st year students of Professional Master in Electronic Commerce at FSJEGJ (Lecture) Period: 2012-2013
* Advanced Data bases
Audience: : 1st year students of Professional Master in Electronic Commerce at FSJEGJ (Lecture) Period: 2013-2014
* Web programming
Audience: 2nd year students of license in computer science FSJEGJ (Lecture) Period: 2012-2013, 2013-2014
Audience: 3rd year students of license in computer science FSJEGJ (Lecture) Period: 2012-2013, 2013-2014, 2014-2015, 2015-2016
* Algorithms and Data Structures I and II
Audience: 1st year students of license in computer science FSJEGJ (TD) Period: 2008-2009, 2009-2010, 2010-2011, 2011-201, 2014-2015 and 2015-2016
* Methodologies of SI conception
Audience: 2nd year students of license in computer science FSJEGJ (TD) Period: 2008-2009, 2009-2010 and 2011-2012
* Data bases
Audience: 1st year students of license in computer science FSJEGJ (TD) Period: 2009-2010, 2010-2011
* Computers architecture
Audience: 2nd year students of license in computer science FSJEGJ (TD) Period: 2008-2009
* Object oriented methodologies for information systems design
Audience: 3rd year students of license in computer science FSJEGJ (TD) Period: 2009-2010, 2010-2011
* Electronic documents exchange (XML)
Audience: 3rd year students of license in computer science FSJEGJ (TD) Period: 2008-2009
* Programming in C
Audience: 1st year students of license in computer science FSJEGJ (Lab works) Period: 2010-2011
* Programming in C++
Audience: 2nd year students of license in computer science FSJEGJ (Lab works) Period: 2010-2011
* Operating systems
Audience: 1st year students of license in computer science FSJEGJ (Lab works) Period: 2013-2014
- 2008-2015: Supervision of licence graduation projects in computer science (50 projects)
- 2012-2015: Co-supervision of Master graduation projects in computer science (6 projects)
- 2013-2015: Supervision of Master graduation projects in E-commerce (7 projects)
- International Journals
Alouaoui, H., Turki, S.Y., and Faiz, S. (2014). “A neural network based on time series for spatiotemporal relationships prediction”, International Journal of Spatial, Temporal and Multimedia Information Systems (accepted paper (in press), 2014)).
Alouaoui, H., Turki, S.Y., and Faiz, S. (2015). “Mining Spatiotemporal Association Rules from Spatiotemporal Databases”, International Journal of Knowledge Engineering and Data Mining, Vol 3, No (2), pp- 190-207.
- International conferences with proceeding
Alouaoui H., Mahmoud A., Turki Y., Roushdy M., Faiz S., Boursier P., Salem A.B. (2013). “Temporal Association Rule mining for Thrombosis Detection”, International Workshop on Artificial Intelligence Technologies for Spatial ￼￼￼￼￼￼￼￼￼￼ Risk Prediction (AITSRP’2013), Le Caire, Egypte, pp. 215-219.
Alouaoui H., Turki S.Y., Faiz S. (2012). “Mining spatiotemporal associations using queries”, Proc. 2nd IEEE International Conference on Information Technology and e-Services (ICITeS’2012), Sousse, Tunisie, Mars 2012, pp. 127-131.
Alouaoui H., Turki S.Y., Faiz S. (2011). “Querying and mining spatiotemporal association rules”, In proceedings of the International Conference on Knowledge Discovery and Information Retrieval (KDIR’2011), Paris, France, pp. 402-405.
Alouaoui H., Turki S.Y., Faiz S. (2010). “Apport de la technique de fouille de données spatiales dans la prédiction des risques engendrés par les changements climatiques”, 10ème conférence internationale sur l’extraction et la gestion des connaissances (EGC’2010), Hammamet, Tunisie.
- International symposiums with program committee
Alouaoui H., Turki S.Y., Faiz S. (2011). “Spatiotemporal datamining for natural risk prediction”, Tunisian Japanese Symposium on Science, Society and Technology (TJASSST’2011), Hammamet, Tunisie.
Training and events
Internship at INSA (Institut National des Sciences Appliqués)- Lyon- France. Subject: Spatiotemporal data mining for risks prediction. Supervisor: Prof. Robert LAURINI. 1-31 July, 2011, Lyon, France.
Member of the Tunisian- Egyptian project "Artificial intelligence for urban risk prediction" September 2012 – January 2014.
( Institutions: ENIT (National engineering school of Tunisia) FCIS-ASU (Faculty of Computer and Information Sciences, Ain Shams University, Egypt) Laboratories: LTSIRS-Tunisia and AIKERU-CSD (AI & Knowledge Engineering Research Unit, Computer Science Dept) - Egypt. Directors: Prof. Sami FAIZ and Dr. Mohamed Roushdy)
Member of the LTSIRS (Laboratory of remote sensing and Spatial Information Systems) National Engineers school (ENIT) - Tunisia, 2010-2015.
ICube Laboratory Télécom Physique Strasbourg 300 bd Sébastien Brant - CS 10413 F - 67412 Illkirch cedex
Office: C320 Phone: +33 (0)3 68 85 44 11 Email: email@example.com