Département informatique

Présentation
Recherche
Enseignement
Doctorants
Conférences
Projets Erasmus+

Les prochains événements du département

Les dernières publications du département
Communication dans un congrès
Mouchet Maxime, Chonavel Thierry, Vaton Sandrine
Statistical Characterization of Round-Trip Times with Nonparametric Hidden Markov Models
IFIP/IEEE IM 2019 Workshop: 4th International Workshop on Analytics for Network and Service Management (AnNet 2019), Apr 2019, Washington, DC, United States
Bibtext :
@inproceedings{mouchet:hal-02010099,
TITLE = {{Statistical Characterization of Round-Trip Times with Nonparametric Hidden Markov Models}},
AUTHOR = {Mouchet, Maxime and Chonavel, Thierry and Vaton, Sandrine},
URL = {https://hal.archives-ouvertes.fr/hal-02010099},
BOOKTITLE = {{IFIP/IEEE IM 2019 Workshop: 4th International Workshop on Analytics for Network and Service Management (AnNet 2019)}},
ADDRESS = {Washington, DC, United States},
YEAR = {2019},
MONTH = Apr,
PDF = {https://hal.archives-ouvertes.fr/hal-02010099/file/AnNet_2019.pdf},
HAL_ID = {hal-02010099},
HAL_VERSION = {v1},
}
Endnote :
%0 Conference Proceedings
%T Statistical Characterization of Round-Trip Times with Nonparametric Hidden Markov Models
%+ Institut de Recherche en Informatique et Systèmes Aléatoires (IRISA)
%+ Département Informatique (IMT Atlantique - INFO)
%+ Université Bretagne Loire (UBL)
%+ Lab-STICC_TB_CID_TOMS
%+ Département Signal et Communications (IMT Atlantique - SC)
%A Mouchet, Maxime
%A Chonavel, Thierry
%A Vaton, Sandrine
%< avec comité de lecture
%B IFIP/IEEE IM 2019 Workshop: 4th International Workshop on Analytics for Network and Service Management (AnNet 2019)
%C Washington, DC, United States
%8 2019-04-08
%D 2019
%Z Computer Science [cs]/Networking and Internet Architecture [cs.NI]Conference papers
%X The study of round-trip time (RTT) measurements on the Internet is of particular importance for improving real-time applications, enforcing QoS with traffic engineering, or detecting unexpected network conditions. On large timescales, from 1 hour to several days, RTT measurements exhibit characteristic patterns due to inter and intra-AS routing changes and traffic engineering, in addition to link congestion. We propose the use of a nonparametric Bayesian model, the Hierarchical Dirichlet Process Hidden Markov Model (HDP-HMM), to characterize RTT timeseries. The parameters of the HMM, including the number of states, as well as the values of hidden states are estimated from delay observations by Gibbs sampling. No assumptions are made on the number of states, and a nonparametric mixture model is used to represent a wide range of delay distribution in each state for more flexibility. We validate the model through three applications: on RIPE Atlas measurements we show that 80% of the states learned on RTTs match only one AS path; on a labelled delay changepoint dataset we show that the model is competitive with state-of-the-art changepoint detection methods in terms of precision and recall; and we show that the predictive ability of the model allows us to reduce the monitoring cost by 90% in routing overlays using Markov decision processes.
%G English
%2 https://hal.archives-ouvertes.fr/hal-02010099/document
%2 https://hal.archives-ouvertes.fr/hal-02010099/file/AnNet_2019.pdf
%L hal-02010099
%U https://hal.archives-ouvertes.fr/hal-02010099
%~ CNRS
%~ UNIV-BREST
%~ UNIV-UBS
%~ INSTITUT-TELECOM
%~ IRISA_SET
%~ ENIB
%~ LAB-STICC_ENIB
%~ LAB-STICC
%~ LAB-STICC_TB
%~ CENTRALESUPELEC
%~ IRISA
%~ INRIA
%~ IMT-ATLANTIQUE
%~ LAB-STICC_IMTA_CID_TOMS
%~ LAB-STICC_IMTA
%~ INRIA-AUT
%~ IMTA_SC
%~ IMTA_INFO
%~ INSA-GROUPE
%~ UNIV-RENNES1
%~ UR1-MATH-STIC
%~ UR1-HAL
%~ UR1-UFR-ISTIC
%~ TEST-UR-CSS
%~ IRISA_IMTA
%~ UNIV-RENNES
%~ PRACOM
Communication dans un congrès
Le Ngoc Luyen, Haralambous Yannis
CCG Supertagging Using Morphological and Dependency Syntax Information
International Conference on Computational Linguistics and Intelligent Text Processing, Apr 2019, La Rochelle, France
Bibtext :
@inproceedings{le:hal-02110790,
TITLE = {{CCG Supertagging Using Morphological and Dependency Syntax Information}},
AUTHOR = {Le, Ngoc Luyen and Haralambous, Yannis},
URL = {https://hal.archives-ouvertes.fr/hal-02110790},
BOOKTITLE = {{International Conference on Computational Linguistics and Intelligent Text Processing}},
ADDRESS = {La Rochelle, France},
HAL_LOCAL_REFERENCE = {19593},
PUBLISHER = {{Springer}},
VOLUME = {Springer Lecture Notes in Computer Science},
PAGES = {.},
YEAR = {2019},
MONTH = Apr,
KEYWORDS = {Dependency Syntax ; Supertagging ; Combinatory Categorial Grammar ; Deep learning},
PDF = {https://hal.archives-ouvertes.fr/hal-02110790/file/CCG_Supertagging%20%282%29.pdf},
HAL_ID = {hal-02110790},
HAL_VERSION = {v1},
}
Endnote :
%0 Conference Proceedings
%T CCG Supertagging Using Morphological and Dependency Syntax Information
%+ Département Informatique (IMT Atlantique - INFO)
%+ Laboratoire des sciences et techniques de l'information, de la communication et de la connaissance (Lab-STICC)
%+ Lab-STICC_IMTA_CID_DECIDE
%A Le, Ngoc Luyen
%A Haralambous, Yannis
%< avec comité de lecture
%Z 19593
%( Proceedings International Conference on Computational Linguistics and Intelligent Text Processing
%B International Conference on Computational Linguistics and Intelligent Text Processing
%C La Rochelle, France
%I Springer
%V Springer Lecture Notes in Computer Science
%P .
%8 2019-04-07
%D 2019
%K Dependency Syntax
%K Supertagging
%K Combinatory Categorial Grammar
%K Deep learning
%Z Computer Science [cs]/Human-Computer Interaction [cs.HC]
%Z Computer Science [cs]/Networking and Internet Architecture [cs.NI]
%Z Computer Science [cs]/Document and Text Processing
%Z Cognitive science/LinguisticsConference papers
%X After presenting a new CCG supertagging algorithm based on morphological and dependency syntax information, we use this algorithm to create a CCG French Tree Bank corpus (20,261 sentences) based on the FTB corpus by Abeillé et al. We then use this corpus, as well as the Groningen Tree Bank corpus for the English language, to train a new BiLSTM+CRF neural architecture that uses (a) morphosyntactic input features and (b) feature correlations as input features. We show experimentally that for an inflected language like French, dependency syntax information allows significant improvement of the accuracy of the CCG supertagging task, when using deep learning techniques.
%G English
%2 https://hal.archives-ouvertes.fr/hal-02110790/document
%2 https://hal.archives-ouvertes.fr/hal-02110790/file/CCG_Supertagging%20%282%29.pdf
%L hal-02110790
%U https://hal.archives-ouvertes.fr/hal-02110790
%~ UNIV-BREST
%~ LAB-STICC_IMTA
%~ IMTA_INFO
%~ LAB-STICC_IMTA_CID_DECIDE
%~ LAB-STICC
%~ ENIB
%~ TELECOM-BRETAGNE
%~ UNIV-UBS
%~ CNRS
%~ INSTITUT-TELECOM
%~ IMT-ATLANTIQUE
Communication dans un congrès
Haralambous Yannis, N'Zi Ehoussou Emmanuel
Salience-induced Term-driven Serendipitous Web Exploration
International Conference on Computational Linguistics and Intelligent Text Processing, Apr 2019, La Rochelle, France
Bibtext :
@inproceedings{haralambous:hal-02121611,
TITLE = {{Salience-induced Term-driven Serendipitous Web Exploration}},
AUTHOR = {Haralambous, Yannis and N'Zi, Ehoussou Emmanuel},
URL = {https://hal.archives-ouvertes.fr/hal-02121611},
BOOKTITLE = {{International Conference on Computational Linguistics and Intelligent Text Processing}},
ADDRESS = {La Rochelle, France},
HAL_LOCAL_REFERENCE = {19592},
PUBLISHER = {{Springer}},
VOLUME = {Springer Lecture Notes in Computer Science},
PAGES = {.},
YEAR = {2019},
MONTH = Apr,
KEYWORDS = {Web exploration ; Serendipity ; Search engine},
PDF = {https://hal.archives-ouvertes.fr/hal-02121611/file/CICLING_2019_Ehoussou_2.pdf},
HAL_ID = {hal-02121611},
HAL_VERSION = {v1},
}
Endnote :
%0 Conference Proceedings
%T Salience-induced Term-driven Serendipitous Web Exploration
%+ Lab-STICC_IMTA_CID_DECIDE
%+ Département Informatique (IMT Atlantique - INFO)
%+ Laboratoire des sciences et techniques de l'information, de la communication et de la connaissance (Lab-STICC)
%A Haralambous, Yannis
%A N'Zi, Ehoussou Emmanuel
%< avec comité de lecture
%Z 19592
%( Proceedings International Conference on Computational Linguistics and Intelligent Text Processing
%B International Conference on Computational Linguistics and Intelligent Text Processing
%C La Rochelle, France
%I Springer
%V Springer Lecture Notes in Computer Science
%P .
%8 2019-04-07
%D 2019
%K Web exploration
%K Serendipity
%K Search engine
%Z Computer Science [cs]/Human-Computer Interaction [cs.HC]
%Z Computer Science [cs]/Networking and Internet Architecture [cs.NI]
%Z Computer Science [cs]/Document and Text Processing
%Z Computer Science [cs]/WebConference papers
%X The Web is a vast place and search engines are our sensory organs to perceive it. To be efficient, Web exploration ideally should have a high serendipity potential. We present a formalization of Web search as a linguistic transformation, evaluate its stability, and apply it to produce serendipity lattices, containing suggestions of term chains to be used as exploration paths. We show experimentally that these lattices conform to two of the three serendipity criteria: relatedness and unexpectedness.
%G English
%2 https://hal.archives-ouvertes.fr/hal-02121611/document
%2 https://hal.archives-ouvertes.fr/hal-02121611/file/CICLING_2019_Ehoussou_2.pdf
%L hal-02121611
%U https://hal.archives-ouvertes.fr/hal-02121611
%~ UNIV-BREST
%~ LAB-STICC_IMTA
%~ IMTA_INFO
%~ LAB-STICC_IMTA_CID_DECIDE
%~ LAB-STICC
%~ ENIB
%~ TELECOM-BRETAGNE
%~ UNIV-UBS
%~ CNRS
%~ INSTITUT-TELECOM
%~ IMT-ATLANTIQUE
Communication dans un congrès
Gilliot Jean-Marie
Vers un environnement de gestion personnelle et collective des données d’apprentissage au service du pouvoir d’agir des apprenants
Séminaire de recherche CIREL : Formation des adultes et Environnements Informatiques pour l'Apprentissage Humain, Apr 2019, Lille, France, Apr 2019, Lille, France
Bibtext :
@inproceedings{gilliot:hal-02086306,
TITLE = {{Vers un environnement de gestion personnelle et collective des donn{\'e}es d'apprentissage au service du pouvoir d'agir des apprenants}},
AUTHOR = {Gilliot, Jean-Marie},
URL = {https://hal.archives-ouvertes.fr/hal-02086306},
BOOKTITLE = {{S{\'e}minaire de recherche CIREL : Formation des adultes et Environnements Informatiques pour l'Apprentissage Humain, Apr 2019, Lille, France}},
ADDRESS = {Lille, France},
YEAR = {2019},
MONTH = Apr,
HAL_ID = {hal-02086306},
HAL_VERSION = {v1},
}
Endnote :
%0 Conference Paper
%F Oral
%T Vers un environnement de gestion personnelle et collective des données d’apprentissage au service du pouvoir d’agir des apprenants
%+ Département Informatique (IMT Atlantique - INFO)
%+ Lab-STICC_IMTA_CID_IHSEV
%A Gilliot, Jean-Marie
%< avec comité de lecture
%B Séminaire de recherche CIREL : Formation des adultes et Environnements Informatiques pour l'Apprentissage Humain, Apr 2019, Lille, France
%C Lille, France
%8 2019-04-01
%D 2019
%Z Computer Science [cs]/Technology for Human LearningConference papers
%G French
%L hal-02086306
%U https://hal.archives-ouvertes.fr/hal-02086306
%~ TICE
%~ CNRS
%~ UNIV-BREST
%~ UNIV-UBS
%~ ENIB
%~ LAB-STICC
%~ LAB-STICC_IMTA_CID_IHSEV
%~ IMT-ATLANTIQUE
%~ IMTA_INFO
%~ INSTITUT-TELECOM
%~ LAB-STICC_IMTA
%~ TEL
Rapport
Luengo Vanda, Guin Nathalie, Bouhineau Denis, Daubias Philippe, Bruillard Eric, Iksal Sébastien, Garlatti Serge, Prié Yannick, Kuzniak Romain
HUBBLE, UN OBSERVATOIRE DES ANALYSES DES TRACES
[Rapport de recherche] ANR (Agence Nationale de la Recherche - France). 2019
Bibtext :
@techreport{luengo:hal-02091392,
TITLE = {{HUBBLE, UN OBSERVATOIRE DES ANALYSES DES TRACES}},
AUTHOR = {Luengo, Vanda and Guin, Nathalie and Bouhineau, Denis and Daubias, Philippe and Bruillard, Eric and Iksal, S{\'e}bastien and Garlatti, Serge and Pri{\'e}, Yannick and Kuzniak, Romain},
URL = {https://hal.archives-ouvertes.fr/hal-02091392},
TYPE = {Research Report},
INSTITUTION = {{ANR (Agence Nationale de la Recherche - France)}},
YEAR = {2019},
MONTH = Mar,
PDF = {https://hal.archives-ouvertes.fr/hal-02091392/file/M%C3%A9moreScientifiqueFinalHubble.pdf},
HAL_ID = {hal-02091392},
HAL_VERSION = {v1},
}
Endnote :
%0 Report
%T HUBBLE, UN OBSERVATOIRE DES ANALYSES DES TRACES
%+ Modèles et Outils en ingénierie des Connaissances pour l'Apprentissage Humain (MOCAH)
%+ Laboratoire d'ingénierie des systèmes d'information (LISI)
%+ MeTAH
%+ Institut français de l'Education (IFÉ)
%+ Sciences Techniques Éducation Formation (STEF)
%+ Laboratoire d'Informatique de l'Université du Maine (LIUM)
%+ Département Informatique (IMT Atlantique - INFO)
%A Luengo, Vanda
%A Guin, Nathalie
%A Bouhineau, Denis
%A Daubias, Philippe
%A Bruillard, Eric
%A Iksal, Sébastien
%A Garlatti, Serge
%A Prié, Yannick
%A Kuzniak, Romain
%I ANR (Agence Nationale de la Recherche - France)
%8 2019-03-31
%D 2019
%Z Computer Science [cs]/Technology for Human LearningReports
%X HUBBLE is organized around two concepts: (1) a study case is a particular learning situation with teachers, learners, managers and a learning platform, (2) an analysis scenario describes the goals of the analysis, the use context and the indicators or models.HUBBLE is organized in an iterative and bottom-up way. The consortium allowed several study cases to be proposed at the beginning of the project and to perform three iterations. The cases share common analysis scenarios: the quality of a MCQ, learners’ typology and the learning pathways. These iterations allowed the analysis scenarios to be improved by using evaluations of study cases for which they were conceived and from new cases for which they were reused; and also to show steps of conception, theirs roles, and to clarify the uses and limits of proposed concepts. Finally, in this bottom-up approach, capitalization, re-utilization and evaluation allowed models of generic e-learning data to be analysed, as well as reusable analysis algorithms and visualizations for use in the field of e-learning.HUBBLE created a catalogue of analysis processes and tools to assist research and the reuse of these processes, and also allowed accessibility to a set of analysis tools. Hubble was responsible for the construction of a national community on the theme of Learning Analytics. Since then, this theme has taken its rightful place in national conferences (Orphée, EIAH), training (masters), collaborations with companies (thesis grants), projects with the Rectorat (Nantes, Orleans-Tours…) supported by the Direction du Numérique pour l’Education (a governmental service managing digital education). At the European level an Erasmus project was applied for in March 2019.The scientific productions are of different kinds: Firstly, concerning tools and methods allowing those not specialized in data analysis or computer science to rapidly explore analysis and traces. In addition, methods of dynamic conception used to propose dashboards suited to the needs of teachers. Next, for the Leaning Analytics community (researchers and analysts) HUBBLE offers the possibility, via formalisms and tools, to capitalize analyses whilst describing them precisely (justification of choices) and independently of an analysis platform, to consult and reuse them. Moreover, the ethical question has been fully integrated into the conception of analysis by taking into account not only data from an ethical point of view, but also in the treatment of analyses.
%X L’objectif d’HUBBLE est de créer un observatoire national pour la construction et le partage de processus d’analyse de données massives, issues des traces des environnements de type e-learning. L’enjeu principal est de permettre la conception, la capitalisation et la réutilisation de ces analyses en associant des acteurs de différents domaines. HUBBLE a vocation à fédérer des équipes de recherche dans plusieurs domaines de l’informatique et des sciences humaines. Le projet considère les problèmes éthiques en proposant un comité éthique et des outils.HUBBLE est organisé autour de deux concepts : (1) un cas d’étude est une situation d’apprentissage particulière, avec des enseignants, des apprenants, des gestionnaires et une plateforme d’apprentissage, (2) un scénario d’analyse décrit les objectifs de l’analyse, le contexte d’usage et les indicateurs ou modèles attendus.Les productions scientifiques du projet sont de plusieurs ordres : D’abord, des outils et méthodes permettant, pour les analystes non spécialisés en analyse de données et non informaticiens, d’explorer rapidement des analyses et des traces ; De plus, des méthodes de conception dynamique afin de proposer des tableaux de bord adaptés aux besoins des enseignants. Ensuite pour la communauté des Learning Analytics (LA), chercheurs et analystes, HUBBLE offre la possibilité, à travers des formalismes et outils, de capitaliser les analyses, en les décrivant de manière fine (justification des choix) et indépendamment d'une plateforme d'analyse, pour ensuite les consulter et les réutiliser. La dimension éthique a été pleinement intégrée dans la conception des analyses en tenant compte non seulement de l’éthique sur les données mais aussi de l’éthique sur les analyses.
%G French
%2 https://hal.archives-ouvertes.fr/hal-02091392/document
%2 https://hal.archives-ouvertes.fr/hal-02091392/file/M%C3%A9moreScientifiqueFinalHubble.pdf
%L hal-02091392
%U https://hal.archives-ouvertes.fr/hal-02091392
%~ TICE
%~ UNIV-GRENOBLE1
%~ UNIV-PMF_GRENOBLE
%~ UNIV-LEMANS
%~ ENS-CACHAN
%~ LIG
%~ UGA
%~ LIUM
%~ SORBONNE-UNIV
%~ ENS-LYON
%~ CNRS
%~ LARA
%~ ENS-CACHAN-SACLAY
%~ INSA-LYON
%~ IMT-ATLANTIQUE
%~ IMTA_INFO
%~ INSA-GROUPE
%~ SORBONNE-UNIVERSITE
%~ LIP6
%~ EIAH
%~ LIG_SIC_METAH
%~ PUBLI-EIAH
%~ TEL
%~ UNIV-PARIS-SACLAY
%~ SU-SCIENCES
Communication dans un congrès
El Ankouri Anas, Neto Luiz, Simon Gaël, Le Bras Hugues, Sanhaji Ali, Chanclou P.
High-Speed Train Cell-less Network Enabled by XGS-PON and Impacts on vRAN Split Interface Transmission
OFC 2019 (Optical Fiber Conference), Mar 2019, San Diego, United States
Bibtext :
@inproceedings{elankouri:hal-02018862,
TITLE = {{High-Speed Train Cell-less Network Enabled by XGS-PON and Impacts on vRAN Split Interface Transmission}},
AUTHOR = {EL ANKOURI, ANAS and Neto, Luiz Anet and Simon, Ga{\"e}l and Le Bras, Hugues and Sanhaji, Ali and Chanclou, P.},
URL = {https://hal-imt-atlantique.archives-ouvertes.fr/hal-02018862},
BOOKTITLE = {{OFC 2019 (Optical Fiber Conference)}},
ADDRESS = {San Diego, United States},
YEAR = {2019},
MONTH = Mar,
PDF = {https://hal-imt-atlantique.archives-ouvertes.fr/hal-02018862/file/OFC%202019.pdf},
HAL_ID = {hal-02018862},
HAL_VERSION = {v1},
}
Endnote :
%0 Conference Paper
%F Oral
%T High-Speed Train Cell-less Network Enabled by XGS-PON and Impacts on vRAN Split Interface Transmission
%+ Orange Labs [Lannion]
%+ Département Informatique (IMT Atlantique - INFO)
%A EL ANKOURI, ANAS
%A Neto, Luiz, Anet
%A Simon, Gaël
%A Le Bras, Hugues
%A Sanhaji, Ali
%A Chanclou, P.
%< avec comité de lecture
%B OFC 2019 (Optical Fiber Conference)
%C San Diego, United States
%8 2019-03-03
%D 2019
%Z 1902.06137
%Z Computer Science [cs]/Networking and Internet Architecture [cs.NI]Conference papers
%X We successfully demonstrate a transmission of a high layer split mobile interface for cell-less, high-speed train network applications using a commercially available XGS-PON. Operation is also demonstrated for a GbE interface.
%G English
%2 https://hal-imt-atlantique.archives-ouvertes.fr/hal-02018862/document
%2 https://hal-imt-atlantique.archives-ouvertes.fr/hal-02018862/file/OFC%202019.pdf
%L hal-02018862
%U https://hal-imt-atlantique.archives-ouvertes.fr/hal-02018862
%~ IMT-ATLANTIQUE
%~ IMTA_INFO
Communication dans un congrès
Manoury Alexandre, Nguyen Sao Mai, Buche Cédric
CHIME: an Adaptive Hierarchical Representation for Continuous Intrinsically Motivated Exploration
The Third IEEE International Conference on Robotic Computing (IRC 2019), Feb 2019, Naples, Italy. pp.167-170
Bibtext :
@inproceedings{manoury:hal-02072393,
TITLE = {{CHIME: an Adaptive Hierarchical Representation for Continuous Intrinsically Motivated Exploration}},
AUTHOR = {Manoury, Alexandre and Nguyen, Sao Mai and Buche, C{\'e}dric},
URL = {https://hal.archives-ouvertes.fr/hal-02072393},
BOOKTITLE = {{The Third IEEE International Conference on Robotic Computing (IRC 2019)}},
ADDRESS = {Naples, Italy},
SERIES = {2019 Third IEEE International Conference on Robotic Computing (IRC)},
PAGES = {167-170},
YEAR = {2019},
MONTH = Feb,
KEYWORDS = {Adaptive ; Continuous learning ; Life-long learning ; Intrinsic motivation ; Goal-babbling ; Hierarchy ; Planning},
PDF = {https://hal.archives-ouvertes.fr/hal-02072393/file/IRC_2019__Short_%20%282%29.pdf},
HAL_ID = {hal-02072393},
HAL_VERSION = {v1},
}
Endnote :
%0 Conference Paper
%F Oral
%T CHIME: an Adaptive Hierarchical Representation for Continuous Intrinsically Motivated Exploration
%+ Département Informatique (IMT Atlantique - INFO)
%+ Lab-STICC_IMTA_CID_IHSEV
%+ Lab-STICC_ENIB_CID_IHSEV
%A Manoury, Alexandre
%A Nguyen, Sao Mai
%A Buche, Cédric
%< avec comité de lecture
%B The Third IEEE International Conference on Robotic Computing (IRC 2019)
%C Naples, Italy
%S 2019 Third IEEE International Conference on Robotic Computing (IRC)
%P 167-170
%8 2019-02-25
%D 2019
%K Adaptive
%K Continuous learning
%K Life-long learning
%K Intrinsic motivation
%K Goal-babbling
%K Hierarchy
%K Planning
%Z Computer Science [cs]/Artificial Intelligence [cs.AI]
%Z Computer Science [cs]/Machine Learning [cs.LG]
%Z Computer Science [cs]/Robotics [cs.RO]Conference papers
%X For a lifelong learning, robots need to learn and adapt to the environment to complete multiple tasks ranging from low-level to high-level tasks, using simple actions or complex ones. Current intrinsically motivated solutions often rely on fixed representations of this environment to define possible tasks, limiting the possibility to adapt to new or changing ones. We propose an algorithm that is able to autonomously 1) self-discover tasks to learn in its environment 2) discover the relationship between tasks to leverage its acquired knowledge on low-level tasks to solve high-level tasks 3) devise a sequence of policies of unbounded length to complete the tasks. Our algorithm, named Continual Hierarchical Intrinsically Motivated Exploration (CHIME), uses planning to build chains of actions, the learning of a hierarchical representation of tasks to reuse low-level skills for high-level tasks and intrinsically-motivated goal babbling to discover new subtasks and orient its learning in its high-dimensional continuous environment. To highlight the features of CHIME, we implement it in a simulated mobile robot where it can move and place objects.
%G English
%2 https://hal.archives-ouvertes.fr/hal-02072393/document
%2 https://hal.archives-ouvertes.fr/hal-02072393/file/IRC_2019__Short_%20%282%29.pdf
%L hal-02072393
%U https://hal.archives-ouvertes.fr/hal-02072393
%~ CNRS
%~ UNIV-BREST
%~ UNIV-UBS
%~ ENIB
%~ LAB-STICC
%~ LAB-STICC_IMTA_CID_IHSEV
%~ IMT-ATLANTIQUE
%~ IMTA_INFO
%~ INSTITUT-TELECOM
%~ LAB-STICC_IMTA
Communication dans un congrès
Saverimoutou Antoine, Mathieu Bertrand, Vaton Sandrine
Web View: Measuring & Monitoring Representative Information on Websites
ICIN 2019 - QOE-MANAGEMENT 2019, Feb 2019, Paris, France
Bibtext :
@inproceedings{saverimoutou:hal-02072471,
TITLE = {{Web View: Measuring \& Monitoring Representative Information on Websites}},
AUTHOR = {Saverimoutou, Antoine and Mathieu, Bertrand and Vaton, Sandrine},
URL = {https://hal.archives-ouvertes.fr/hal-02072471},
BOOKTITLE = {{ICIN 2019 - QOE-MANAGEMENT 2019}},
ADDRESS = {Paris, France},
YEAR = {2019},
MONTH = Feb,
KEYWORDS = {Web Browsing ; Web Monitoring ; QoE ; QoS},
PDF = {https://hal.archives-ouvertes.fr/hal-02072471/file/WebView-QoE_Management2019.pdf},
HAL_ID = {hal-02072471},
HAL_VERSION = {v1},
}
Endnote :
%0 Conference Paper
%F Oral
%T Web View: Measuring & Monitoring Representative Information on Websites
%+ Orange Labs [Lannion]
%+ Département Informatique (IMT Atlantique - INFO)
%+ Advanced technologies for operated networks (ADOPNET)
%+ IMT Atlantique Bretagne-Pays de la Loire (IMT Atlantique)
%+ RÉSEAUX, TÉLÉCOMMUNICATION ET SERVICES (IRISA_D2)
%A Saverimoutou, Antoine
%A Mathieu, Bertrand
%A Vaton, Sandrine
%< avec comité de lecture
%B ICIN 2019 - QOE-MANAGEMENT 2019
%C Paris, France
%8 2019-02-18
%D 2019
%K Web Browsing
%K Web Monitoring
%K QoE
%K QoS
%Z Computer Science [cs]/Networking and Internet Architecture [cs.NI]Conference papers
%X Web browsing is a fast-paced and changing domain which delivers a wide set of services to end-users. While end-users' QoE (Quality of Experience) is the main focus of large service companies and service providers, websites' structure on the one hand change relentlessly. There is therefore a strong need to measure the temporal quality of web browsing by using user-representative web browsers, different Internet protocols and types of residential network access. We present in this paper our tool, Web view, which performs automated desk-top web browsing sessions where each measurement provides 61 monitoring parameters. An analysis of 8 Million different measurements of web pages' homepage indicate which and how different parameters can impact today's web browsing quality. Web measurement campaigns conducted since February 2018 are represented through a monitoring website. Our analysis and monitoring of measurements on a 10-month period show that Google Chrome brings increased web browsing quality.
%G English
%2 https://hal.archives-ouvertes.fr/hal-02072471/document
%2 https://hal.archives-ouvertes.fr/hal-02072471/file/WebView-QoE_Management2019.pdf
%L hal-02072471
%U https://hal.archives-ouvertes.fr/hal-02072471
%~ CNRS
%~ UNIV-UBS
%~ IRISA_SET
%~ UNIV-RENNES1
%~ UR1-HAL
%~ CENTRALESUPELEC
%~ TEST-IRISA-TB
%~ IRISA
%~ INRIA
%~ IMT-ATLANTIQUE
%~ TEST-UR-CSS
%~ UNIV-RENNES
%~ IRISA-ADOPNET-IMTA
%~ IMTA_INFO
%~ INSA-GROUPE
%~ UR1-MATH-STIC
%~ UR1-UFR-ISTIC
%~ IRISA_IMTA
%~ INRIA-AUT
Communication dans un congrès
Fahed Lina, Lenca Philippe, Haralambous Yannis, Lefort Riwal, Tallec Marie-Laure
Prédiction d'évènements distants basée sur les règles séquentielles
EGC 2019 : Extraction et gestion des connaissances, Jan 2019, Metz, France
Bibtext :
@inproceedings{fahed:hal-01962067,
TITLE = {{Distant events prediction based on sequential rules}},
AUTHOR = {Fahed, Lina and Lenca, Philippe and Haralambous, Yannis and Lefort, Riwal and Tallec, Marie-Laure},
URL = {https://hal.archives-ouvertes.fr/hal-01962067},
BOOKTITLE = {{EGC 2019 : Extraction et gestion des connaissances}},
ADDRESS = {Metz, France},
HAL_LOCAL_REFERENCE = {19290},
PAGES = {.},
YEAR = {2019},
MONTH = Jan,
PDF = {https://hal.archives-ouvertes.fr/hal-01962067/file/egc19_shortVersion.pdf},
HAL_ID = {hal-01962067},
HAL_VERSION = {v1},
}
Endnote :
%0 Conference Paper
%F Oral
%T Distant events prediction based on sequential rules
%+ Département Informatique (IMT Atlantique - INFO)
%+ Laboratoire des sciences et techniques de l'information, de la communication et de la connaissance (Lab-STICC)
%+ Lab-STICC_IMTA_CID_DECIDE
%+ Département Logique des Usages, Sciences sociales et Sciences de l'Information (IMT Atlantique - LUSSI)
%+ Arkéa
%A Fahed, Lina
%A Lenca, Philippe
%A Haralambous, Yannis
%A Lefort, Riwal
%A Tallec, Marie-Laure
%< avec comité de lecture
%Z 19290
%3 Actes EGC 2019 : Extraction et gestion des connaissances
%B EGC 2019 : Extraction et gestion des connaissances
%C Metz, France
%P .
%8 2019-01-21
%D 2019
%Z Computer Science [cs]/Machine Learning [cs.LG]
%Z Computer Science [cs]/Artificial Intelligence [cs.AI]
%Z Computer Science [cs]/Human-Computer Interaction [cs.HC]
%Z Computer Science [cs]/Computers and Society [cs.CY]
%Z Computer Science [cs]/Networking and Internet Architecture [cs.NI]
%Z Statistics [stat]/Machine Learning [stat.ML]Conference papers
%X Events prediction in sequences database is a challenging task that can be approached with data mining. In this paper, we focus on the specific goal of the prediction of distant events by mining sequential rules and propose the algorithm D-SR (Distant Sequential Rules). The originality of our algorithm is that it mines sequential rules with a consequent temporally distant from the antecedent by applying a minimal gap constraint between both elements. Our algorithm can be integrated into traditional algorithms as a post-processing step or during the mining process, with low time and memory complexity. Experiments show that D-SR is more efficient, in terms of running time and scalability, in mining distant rules during the mining process than in a post- processing step. By studying in details the rules mined from a real-world dataset, we demonstrate the efficiency of such rules for real-world applications.
%X La prédiction d'évènements dans une base de séquences est une tâche difficile à laquelle la fouille de données peut largement contribuer. Dans cet article nous nous concentrons sur le cas spécifique de la prédiction d'évènements distants à travers la fouille de règles séquentielles and nous proposons l'algorithme D-SR (règles séquentielles distantes). L'originalité de notre algorithme réside dans le fait qu'il fouille les règles séquentielles avec un conséquent temporellement distant de l'antécédent, en appliquant une contrainte de gap minimal entre les deux éléments. Notre algorithme peut être intégré dans les algorithmes traditionnels en tant qu'étape de post-traitement ou pendant le processus de fouille, avec une faible complexité de temps et de mémoire. Nos expériences montrent que D-SR est plus efficace, en termes de temps de calcul et de scalabilité, quand on fouille des règles distantes pendant le processus de fouille que plutôt en tant qu'étape de post-traitement. En étudiant en détail les règles obtenues par la fouille d'un ensemble de données, nous montrons leur efficacité pour des applications du monde réel.
%G French
%2 https://hal.archives-ouvertes.fr/hal-01962067/document
%2 https://hal.archives-ouvertes.fr/hal-01962067/file/egc19_shortVersion.pdf
%L hal-01962067
%U https://hal.archives-ouvertes.fr/hal-01962067
%~ UNIV-BREST
%~ LAB-STICC_IMTA
%~ IMTA_LUSSI
%~ IMTA_INFO
%~ LAB-STICC_IMTA_CID_DECIDE
%~ TDS-MACS
%~ LAB-STICC
%~ ENIB
%~ TELECOM-BRETAGNE
%~ UNIV-UBS
%~ CNRS
%~ INSTITUT-TELECOM
%~ IMT-ATLANTIQUE
Article dans une revue
Duminy Nicolas, Nguyen Sao Mai, Duhaut Dominique
Learning a set of interrelated tasks by using a succession of motor policies for a socially guided intrinsically motivated learner
Frontiers in Neurorobotics, Frontiers, 2019, ⟨10.3389/fnbot.2018.00087⟩
Bibtext :
@article{duminy:hal-02087995,
TITLE = {{Learning a set of interrelated tasks by using a succession of motor policies for a socially guided intrinsically motivated learner}},
AUTHOR = {Duminy, Nicolas and Nguyen, Sao Mai and Duhaut, Dominique},
URL = {https://hal.archives-ouvertes.fr/hal-02087995},
JOURNAL = {{Frontiers in Neurorobotics}},
PUBLISHER = {{Frontiers}},
YEAR = {2019},
DOI = {10.3389/fnbot.2018.00087},
KEYWORDS = {intrinsic motivation ; goal-babbling ; multi-task learning ; interactive learning ; active learning ; active imitation learning ; hierarchical learning ; procedures},
PDF = {https://hal.archives-ouvertes.fr/hal-02087995/file/Frontiers2019.pdf},
HAL_ID = {hal-02087995},
HAL_VERSION = {v1},
}
Endnote :
%0 Journal Article
%T Learning a set of interrelated tasks by using a succession of motor policies for a socially guided intrinsically motivated learner
%+ Lab-STICC_TB_CID_IHSEV
%+ Université Bretagne Loire (UBL)
%+ Laboratoire des sciences et techniques de l'information, de la communication et de la connaissance (Lab-STICC)
%+ Université de Bretagne Sud (UBS)
%+ Département Informatique (IMT Atlantique - INFO)
%+ Lab-STICC_IMTA_CID_IHSEV
%+ Laboratoire des sciences et techniques de l'information, de la communication et de la connaissance (Lab-STICC)
%A Duminy, Nicolas
%A Nguyen, Sao Mai
%A Duhaut, Dominique
%< avec comité de lecture
%@ 1662-5218
%J Frontiers in Neurorobotics
%I Frontiers
%8 2019
%D 2019
%R 10.3389/fnbot.2018.00087
%K intrinsic motivation
%K goal-babbling
%K multi-task learning
%K interactive learning
%K active learning
%K active imitation learning
%K hierarchical learning
%K procedures
%Z Computer Science [cs]/Artificial Intelligence [cs.AI]
%Z Computer Science [cs]/Robotics [cs.RO]
%Z Computer Science [cs]/Human-Computer Interaction [cs.HC]
%Z Statistics [stat]/Machine Learning [stat.ML]Journal articles
%X We aim at a robot capable to learn sequences of actions to achieve a field of complex tasks. In this paper, we are considering the learning of a set of interrelated complex tasks hierarchically organized. To learn this high-dimensional mapping between a continuous high-dimensional space of tasks and an infinite dimensional space of unbounded sequences of actions, we introduce a new framework called "procedures", which enables the autonomous discovery of how to combine previously learned skills in order to learn increasingly complex combinations of motor policies. We propose an active learning algorithmic architecture, capable of organizing its learning process in order to achieve a field of complex tasks by learning sequences of primitive motor policies. Based on heuristics of active imitation learning, goal-babbling and strategic learning using intrinsic motivation, our algorithmic architecture leverages our procedures framework to actively decide during its learning process which outcome to focus on and which exploration strategy to apply. We show on a simulated environment that our new architecture is capable of tackling the learning of complex motor policies by adapting the complexity of its policies to the task at hand. We also show that our "procedures" enable the learning agent to discover the task hierarchy and exploit his experience of previously learned skills to learn new complex tasks.
%G English
%2 https://hal.archives-ouvertes.fr/hal-02087995/document
%2 https://hal.archives-ouvertes.fr/hal-02087995/file/Frontiers2019.pdf
%L hal-02087995
%U https://hal.archives-ouvertes.fr/hal-02087995
%~ CNRS
%~ UNIV-BREST
%~ UNIV-UBS
%~ ENIB
%~ LAB-STICC_ENIB
%~ LAB-STICC
%~ LAB-STICC_IMTA_CID_IHSEV
%~ LAB-STICC_TB
%~ IMT-ATLANTIQUE
%~ IMTA_INFO
%~ INSTITUT-TELECOM
%~ LAB-STICC_IMTA
Voir toutes les publications du département
Contacts
Responsable de département : Serge GARLATTI
Tél : +33 2 29 00 14 53
Assistante : Armelle LANNUZEL
Tél : +33 2 29 00 14 06
Voir tous les membres du département
Infos pratiques :
Accès au département Informatique : il est situé dans le bâtiment D3.