Éditeur(s) : HAL CCSD ACM New York, NY, USA
Résumé : International audience
This paper focuses on the study of the behavior of a genetic algorithm based classifier system, the Adapted Pittsburgh Classifier System (A.P.C.S), on maze type environments containing aliasing squares. This type of environment is often used in reinforcement learning litterature to assess the performances of learning methods when facing problems containing non markov situations. Through this study, we discuss on the performances of the APCS on maze type environments and also of the efficiency of an improvement of the APCS learning method inspired from the XCS : the covering mechanism. We manage to show that, without any memory mechanism, the APCS is able to build and to keep accurate strategies to produce regular sub-optimal solutions to these maze problem. This statement is shown through a comparison of the results obtained by the XCS, XCSM and XCSMH on distinct maze problems with these obtained by the APCS.
Proceedings of the 2008 GECCO conference companion on Genetic and evolutionary computation
Atlanta, GA, United States

hal-00520607
https://hal.archives-ouvertes.fr/hal-00520607
DOI : 10.1145/1388969.1389013

Éditeur(s) : HAL CCSD springer
Résumé : International audience
This paper focuses on the study of the behavior of a genetic algorithm based classier system, the Adapted Pittsburgh Classier System (A.P.C.S), on maze type environments containing aliasing squares. This type of environment is often used in reinforcement learning literature to assess the performances of learning methods when facing problems containing non markovian situations. Through this study, we discuss on the performance of the APCS upon two mazes (Woods 101 and Maze E2) and also on the eciency of an improvement of the APCS learning method inspired from the XCS: the covering mechanism. We manage to show that, without any memory mechanism, the APCS is able to build and to keep accurate strategies to produce regular sub-optimal solution to these maze problems. This statement is shown through a comparison between the results obtained by the XCS on two specic maze problems and those obtained by the APCS.
Lecture notes in computer science

hal-00542922
https://hal.archives-ouvertes.fr/hal-00542922
DOI : 10.1007/978-3-642-17508-4_8

Éditeur(s) : HAL CCSD
Résumé : LPSC14037
Scientific data collected with modern sensors or dedicated detectors exceed very often the perimeter of the initial scientific design. These data are obtained more and more frequently with large material and human efforts. A large class of scientific experiments are in fact unique because of their large scale, with very small chances to be repeated and to superseded by new experiments in the same domain: for instance high energy physics and astrophysics experiments involve multi-annual developments and a simple duplication of efforts in order to reproduce old data is simply not affordable. Other scientific experiments are in fact unique by nature: earth science, medical sciences etc. since the collected data is "time-stamped" and thereby non-reproducible by new experiments or observations. In addition, scientific data collection increased dramatically in the recent years, participating to the so-called "data deluge" and inviting for common reflection in the context of "big data" investigations. The new knowledge obtained using these data should be preserved long term such that the access and the re-use are made possible and lead to an enhancement of the initial investment. Data observatories, based on open access policies and coupled with multi-disciplinary techniques for indexing and mining may lead to truly new paradigms in science. It is therefore of outmost importance to pursue a coherent and vigorous approach to preserve the scientific data at long term. The preservation remains nevertheless a challenge due to the complexity of the data structure, the fragility of the custom-made software environments as well as the lack of rigorous approaches in workflows and algorithms. To address this challenge, the PREDON project has been initiated in France in 2012 within the MASTODONS program: a Big Data scientific challenge, initiated and supported by the Interdisciplinary Mission of the National Centre for Scientific Research (CNRS). PREDON is a study group formed by researchers from different disciplines and institutes. Several meetings and workshops lead to a rich exchange in ideas, paradigms and methods. The present document includes contributions of the participants to the PREDON Study Group, as well as invited papers, related to the scientific case, methodology and technology. This document should be read as a "facts finding" resource pointing to a concrete and significant scientific interest for long term research data preservation, as well as to cutting edge methods and technologies to achieve this goal. A sustained, coherent and long term action in the area of scientific data preservation would be highly beneficial.
http://hal.in2p3.fr/in2p3-00959072

in2p3-00959072
http://hal.in2p3.fr/in2p3-00959072
http://hal.in2p3.fr/in2p3-00959072/document
http://hal.in2p3.fr/in2p3-00959072/file/PREDON_Scientific_Data_Preservation_2014.pdf

Éditeur(s) : HAL CCSD ACM
Résumé : International audience
In this study, we use an adapted version of Pitsburgh-like learning classifier system to perform over classification tasks. The Adapted Pittsburgh Classifier System, enhanced with a new mechanism, allows us to consider the classification problems and their treatment by a given LCS in a different manner. Our aim is to exhibit elements in order to prove that, using the action covering mechanism, this system is able to build an inner map of a given classification learning sample. In the context of this paper, this map is built using an intrisic property of Pittsburgh-like CS: the use of various collections of classifiers amongst a unique population.
Proceedings of the 12th annual conference comp on Genetic and evolutionary computation
Portland,Oregon, United States

hal-00520609
https://hal.archives-ouvertes.fr/hal-00520609
DOI : 10.1145/1830761.1830823

Éditeur(s) : HAL CCSD
Résumé : International audience
This paper focuses on the study of the influence of a newly implemented mechanism on a Pittsburgh-like classifier system. The Adapted Pittsburgh Classifier System is a learning classifier system that uses genetic algorithms to evolve its ruleset. The new mechanism discussed is inspired from Wilson work on the eXtended Classifier System (XCS): it allows the concerned LCS to adapt its rule set when facing a new signal by modifying an existing rule. This mechanism is called covering mechanism due to the fact that the rule is covered up by a new rule which is sensitive to this new signal. Effects of this covering mechanism are first measured on the performances of APCS on well known multi-step environments: maze-type environments (Woods 101 and E2). In addition, further measures presented in this paper indicate that we could possibly rely on the behavior of this covering mechanism to automate the research of a correct cognitive capacity needed by the APCS to solve a given multistep problem.
2008 Eighth International Conference on Hybrid Intelligent Systems
Barcelone, Spain

hal-00520608
https://hal.archives-ouvertes.fr/hal-00520608
DOI : 10.1109/HIS.2008.105

Éditeur(s) : HAL CCSD Elsevier
Résumé : International audience
Nitrogen (N) transfer from legume trees to associated crops is a key factor for the N economy of low-input agroforestry systems. In this work, we presented a new approach to estimate N transfer based on the 15N content of root exudates and N released by root turnover of the donor plant (Gliricidia sepium) and the temporal change of the 15N content of the receiver plant (Dichanthium aristatum). The study was carried out in greenhouse using two isotopic methods: 15N leaf feeding (LF) and the natural 15N abundance (NA). Measurements of exudate 15N were made at several dates before and after tree pruning. A time-dependent box model was devised to quantify N transfer in time and to make comparisons between the isotopic methods. In NA, although tree roots and exudates presented a similar 15N signature before tree pruning, exudates were strongly depleted in 15N after pruning. In LF, exudates were always depleted in 15N in relation to tree roots. Hence, the current assumption used in N transfer studies concerning the equal 15N/14N distribution in tissues of the donor plant and in its excreted N was not confirmed in our study. Before pruning, N transfer functioned as a two-N-source system (soil N and exudates N) and both isotopic methods provided similar estimates: 11–12% for LF and 10–15% for NA. Calculations performed with the model indicated that N transfer occurred with small or nil fractionation of 15N in exudates. After pruning, there was a third N source associated with N released from tree root turnover. During this period, the isotopic signature of the receiver plant showed a transient state due to the progressive decrease of 15N content of that N source. The amount of N derived from the tree represented 65% of the total N content of the grass at the end of the experiments.
ISSN: 1161-0301

hal-01082183
https://hal-agroparistech.archives-ouvertes.fr/hal-01082183
DOI : 10.1016/j.eja.2007.03.003

Éditeur(s) : Lugdun. Batavorum : apud Franciscum Hackium Amsterdam : Elzevier
Résumé : "Titre gravé sur cuivre (f. *1) suivi du faux-titre imprimé (f. *2) : "Guilielmi Pisonis, M. D. Lugduno-Batavi, de medicina Brasiliensi libri quatuor: I. De aëre, aquis, & locis. II. De morbis endemiis. III. De venenatis & antidotis. IV. De facultatibus simplicium. Et Georgii Marcgravii de Liebstad, misnici germani, historiæ rerum naturalium Brasiliæ, libri octo: quorum tres priores agunt de plantis. Quartus de piscibus. Quintus de avibus. Sextus de quadrupedibus & serpentibus. Septimus de insectis. Octavus de ipsa regione, & illius incolis. Cum appendice de Tapuyis, et Chilensibus. Joannes de Laet, Antwerpianus, in ordinem digessit & annotationes addidit, & varia ab auctore omissa supplevit & illustravit". - La 2e partie est précédée du faux-titre : "Georgii Marcgravii de Liebstad, misnici germani, historiæ rerum naturalium Brasiliæ, libri octo: quorum tres priores agunt de plantis. Quartus de piscibus. Quintus de avibus. Sextus de quadrupedibus, et serpentibus. Septimus de insectis. Octavus de ipsa regione, & illius incolis. Cum appendice de Tapuyis, et Chilensibus. Joannes de Laet, Antwerpianus, in ordinem digessit & annotationes addidit multas, & varia ab auctore omissa supplevit & illustravit". - Épître dédicatoire au comte Maurice de Nassau Edition originale. - Frontispice gravé sur cuivre par Theodor Matham. - 436 bois dans le texte, d'après Georg Marcgrave et Albert Eckhout Imprimé à Leyde par Bonaventure et Abraham Elzevier, d'après Rahir. Sig. *6, A-P4, Q2 ; -4, A-Oo4, Pp2. - Demi reliure basane, 19e siècle. - Cachet de l'Académie royale de Bordeaux. - Cachet de la bibliothèque de la Ville de Bordeaux Borba de Moraes II, 675-6. - Garrison-Morton 5303. - Nissen BBI 1553. - Nissen IVB 589. - Rahir 1077. - Sabin 63028. - Willems 1068"
Siècle(s) traité(s) : 17
Droits : Domaine public
Provenance : Ville de Bordeaux. Bibliothèque municipale
Permalien : http://www.manioc.org/patrimon/BBX17010
BBX17010

Date de création : 1869-01-01

Éditeur(s) : London : Bell & Daldy
Siècle(s) traité(s) : 19
Droits : Domaine public
Provenance : Collectivité territoriale de Guyane. Bibliothèque Alexandre-Franconie
Permalien : http://www.manioc.org/patrimon/FRA11001
FRA11001

Éditeur(s) : Paris : Firmin Didot Frères

Droits : Domaine public
Provenance : Archives territoriales de Guyane
Permalien : http://www.manioc.org/patrimon/GAD12001
GAD12001

Éditeur(s) : Paris : Firmin Didot Frères

Droits : Domaine public
Provenance : Archives territoriales de Guyane
Permalien : http://www.manioc.org/patrimon/GAD12002
GAD12002