Éditeur(s) : HAL CCSD Presses Universitaires de Louvain (distribution I6DOC/CIACO)
Résumé : In French. 10 pages, 5 figures, LaTeX 2e using EPSF and custom package taln2006.sty (designed by Pierre Zweigenbaum, ATALA). Proceedings of the 13th annual French-speaking conference on Natural Language Processing: `Traitement Automatique des Langues Naturelles' (TALN 2006), Louvain (Leuven), Belgium, 10-13 April 2003
National audience
We propose a theoretical framework within which information on the vocabulary of a given corpus can be inferred on the basis of statistical information gathered on that corpus. Inferences can be made on the categories of the words in the vocabulary, and on their syntactical properties within particular languages. Based on the same statistical data, it is possible to build matrices of syntagmatic similarity (bigram transition matrices) or paradigmatic similarity (probability for any pair of words to share common contexts). When clustered with respect to their syntagmatic similarity, words tend to group into sublanguage vocabularies, and when clustered with respect to their paradigmatic similarity, into syntactic or semantic classes. Experiments have explored the first of these two possibilities. Their results are interpreted in the frame of a Markov chain modelling of the corpus' generative processe(s): we show that the results of a spectral analysis of the transition matrix can be interpreted as probability distributions of words within clusters. This method yields a soft clustering of the vocabulary into sublanguages which contribute to the generation of heterogeneous corpora. As an application, we show how multilingual texts can be visually segmented into linguistically homogeneous segments. Our method is specifically useful in the case of related languages which happened to be mixed in corpora.
Verbum ex machina : Actes de la 13ème conférence annuelle sur le Traitement Automatique des Langues Naturelles
Louvain (Leuven), Belgium

hal-00327607
https://hal.archives-ouvertes.fr/hal-00327607
ARXIV : 0810.1212

Éditeur(s) : HAL CCSD IOS Press (Amsterdam)
Résumé : 6 pages, 7 embedded figures, LaTeX 2e using the ecai2006.cls document class and the algorithm2e.sty style file (+ standard packages like epsfig, amsmath, amssymb, amsfonts...). Extends the short version contained in the ECAI 2006 proceedings.
International audience
Without prior knowledge, distinguishing different languages may be a hard task, especially when their borders are permeable. We develop an extension of spectral clustering -- a powerful unsupervised classification toolbox -- that is shown to resolve accurately the task of soft language distinction. At the heart of our approach, we replace the usual hard membership assignment of spectral clustering by a soft, probabilistic assignment, which also presents the advantage to bypass a well-known complexity bottleneck of the method. Furthermore, our approach relies on a novel, convenient construction of a Markov chain out of a corpus. Extensive experiments with a readily available system clearly display the potential of the method, which brings a visually appealing soft distinction of languages that may define altogether a whole corpus.
ECAI 2006: 17th European Conference on Artificial Intelligence
Riva del Garda, Italy

hal-00327782
https://hal.archives-ouvertes.fr/hal-00327782
ARXIV : 0810.1261

Éditeur(s) : HAL CCSD
Résumé : International audience
The study of complex biological processes requires to forgo simplified models for extensive ones. Yet, these models' size and complexity place them beyond understanding. Their analysis requires new methods for identifying general patterns. The Transforming Growth Factor TGF-β is a multifunctional cytokine that regulates mammalian cell development , differentiation, and homeostasis. Depending on the context, it can play the antagonistic roles of growth inhibitor or of tumor promoter. Its context-dependent pleiotropic nature is associated with complex sig-naling pathways. The most comprehensive model of TGF-β-dependent signaling is composed of 15,934 chains of reactions (trajectories) linking TGF-β to at least one of its 159 target genes. Identifying functional patterns in such a network requires new automated methods. This article presents a framework for identifying groups of similar trajec-tories composed of the same molecules using an exhaustive and without prior assumptions approach. First, the trajectories were clustered using the Relevant Set Correlation model, a shared nearest-neighbors clustering method. Five groups of trajectories were identified. Second, for each cluster the over-represented molecules were determined by scoring the frequency of each molecule implicated in trajectories. Third, Gene set enrichment analysis on the clusters of trajectories revealed some specific TGF-β-dependent biological processes, with different clusters associated to the antagonists roles of TGF-β. This confirms that our approach yields biologically-relevant results. We developed a web interface that facilitates graph visualization and analysis. Our clustering-based method is suitable for identifying families of functionally-similar trajectories in the TGF-β signaling network. It can be generalized to explore any large-scale biological pathways.
CMSB 2017 - 15th International Conference on Computational Methods in Systems Biology
Darmstadt, France

hal-01559249
https://hal.archives-ouvertes.fr/hal-01559249
https://hal.archives-ouvertes.fr/hal-01559249/document
https://hal.archives-ouvertes.fr/hal-01559249/file/JeanCoquet_CMSB.pdf

Éditeur(s) : HAL CCSD Elsevier
Résumé : International audience
Recently, a large amount of work has been devoted to the study of spectral clustering — a powerful unsupervised classification method. This paper brings contributions to both its foundations, and its applications to text classification. Departing from the mainstream, concerned with hard membership, we study the extension of spectral clustering to soft membership (probabilistic, EM style) assignments. One of its key features is to avoid the complexity gap of hard membership. We apply this theory to a challenging problem, text clustering for languages having permeable borders, via a novel construction of Markov chains from corpora. Experiments with a readily available code clearly display the potential of the method, which brings a visually appealing soft distinction of languages that may define altogether a whole corpus.
ISSN: 0031-3203

hal-00329991
https://hal.archives-ouvertes.fr/hal-00329991
DOI : 10.1016/j.patcog.2008.06.024

Éditeur(s) : HAL CCSD
Résumé : One of the major challenges in forest ecology is to understand the ecology, physiology and potential responses of the> 250000 species of vascular plants of the planet through the categorization of this diversity into functional groups. This approach is essentially based on functional traits analysis, but few investigations were made in tropical areas. The aim of this study was to determine functional groups in two contrasted tropical ecosystems, dry (DF) and mountain rain (MRF) forest (respectively rainfall of 1 200 mm/year and 10 000 mm/year). Guadeloupe Island in FWI presents both these ecosystems spread over less 100 km. In a first time, we developed an original species typology based on 7 "soft" traits, easy to measure (stomatal density, leaf trichom density, chlorophyll content, leaf nitrogen content, specific leaf area, wood density, leaf area, plant height). The method was based on principal component analysis (PCA) and agglomerative hierarchical clustering (AHC) applied to these traits measured at the individual level (Blazy et al., 2009). In a second time, we evaluated the relevance of these soft trait-based groups compared to a typology based on 5 "hard" ecophysiological traits, more costly to quantify but with higher functional significance (difference between predawn and minimum leaf water potential, stomatal conductance, maximal carbon assimilation rate, leaf water potential at 90% of loss of stem hydraulic conductivity). When performed on the dominant canopy species in DF and MRF (a total of 13 species and 60 mature individuals), the typological approach based on soft traits seems relevant to discriminate groups with a distinct functional reality. Through both ecosystems, we obtained 4 homogenous groups. Two were exclusively composed by DF's or MRF's species, indicating that a functional strategy can be linked to specific environmental conditions. The two other groups showed a mix of DF and MRF species indicating that different functional strategies can coexist in the same environment. These mixed groups seem distinct by their ecophysiological features, particularly in term of hydraulic characteristics. One group was composed by drought tolerant species, water spender species, with a high xylem embolism resistance, whereas the second one was composed by water saver species, less resistant to embolism. Therefore, a determination of functional groups based on soft traits could be useful to model ecosystem's responses to environmental changes on a global scale.
Book of Abstracts - isem 2013 - Ecological Modelling for Ecosystem Sustainability
Toulouse, France

hal-00964668
https://hal.archives-ouvertes.fr/hal-00964668

Éditeur(s) : HAL CCSD Wiley
Résumé : International audience
Mechanical forces influence the growth and shape of virtually all tissues and organs. Recent studies show that increased cell contractibility, growth and differentiation might be normalized by modulating cell tensions. Particularly, the role of these tensions applied by the extracellular matrix during liver fibrosis could influence the hepatocarcinogenesis process. The objective of this study is to determine if 3D stiffness could influence growth and phenotype of normal and transformed hepatocytes and to integrate extracellular matrix (ECM) stiffness to tensional homeostasis. We have developed an appropriate 3D culture model: hepatic cells within three-dimensional collagen matrices with varying rigidity. Our results demonstrate that the rigidity influenced the cell phenotype and induced spheroid clusters development whereas in soft matrices, Huh7 transformed cells were less proliferative, well-spread and flattened. We confirmed that ERK1 played a predominant role over ERK2 in cisplatin-induced death, whereas ERK2 mainly controlled proliferation. As compared to 2D culture, 3D cultures are associated with epithelial markers expression. Interestingly, proliferation of normal hepatocytes was also induced in rigid gels. Furthermore, biotransformation activities are increased in 3D gels, where CYP1A2 enzyme can be highly induced/activated in primary culture of human hepatocytes embedded in the matrix. In conclusion, we demonstrated that increasing 3D rigidity could promote proliferation and spheroid developments of liver cells demonstrating that 3D collagen gels are an attractive tool for studying rigidity-dependent homeostasis of the liver cells embedded in the matrix and should be privileged for both chronic toxicological and pharmacological drug screening.
ISSN: 0730-2312

hal-01196600
https://hal-univ-rennes1.archives-ouvertes.fr/hal-01196600
https://hal-univ-rennes1.archives-ouvertes.fr/hal-01196600/document
https://hal-univ-rennes1.archives-ouvertes.fr/hal-01196600/file/Increasing%203D%20Matrix%20Rigidity_accepted.pdf
DOI : 10.1002/jcb.25356