Apprentissage automatique de la morphologie : le cas des structures racine-schème

RÉSUMÉ Cette thèse porte sur le développement de méthodes algorithmiques pour découvrir automatiquement la structure morphologique des mots d'un corpus. On considère en particulier le cas des langues s'approchant du type introflexionnel, comme l'arabe ou l'hébreu. La tradition linguistique décrit la morphologie de ces langues en termes d'unités discontinues : les racines consonantiques et les schèmes vocaliques. Ce genre de structure constitue un défi pour les systèmes actuels d'apprentissage automatique, qui opèrent généralement avec des unités continues. La stratégie adoptée ici consiste à traiter le problème comme une séquence de deux sous-problèmes. Le premier est d'ordre phonologique : il s'agit de diviser les symboles (phonèmes, lettres) du corpus en deux groupes correspondant autant que possible aux consonnes et voyelles phonétiques. Le second est de nature morphologique et repose sur les résultats du premier : il s'agit d'établir l'inventaire des racines et schèmes du corpus et de déterminer leurs règles de combinaison. On examine la portée et les limites d'une approche basée sur deux hypothèses : (i) la distinction entre consonnes et voyelles peut être inférée sur la base de leur tendance à alterner dans la chaîne parlée; (ii) les racines et les schèmes peuvent être identifiés respectivement aux séquences de consonnes et voyelles découvertes précédemment. L'algorithme proposé utilise une méthode purement distributionnelle pour partitionner les symboles du corpus. Puis il applique des principes analogiques pour identifier un ensemble de candidats sérieux au titre de racine ou de schème, et pour élargir progressivement cet ensemble. Cette extension est soumise à une procédure d'évaluation basée sur le principe de la longueur de description minimale, dans- l'esprit de LINGUISTICA (Goldsmith, 2001). L'algorithme est implémenté sous la forme d'un programme informatique nommé ARABICA, et évalué sur un corpus de noms arabes, du point de vue de sa capacité à décrire le système du pluriel. Cette étude montre que des structures linguistiques complexes peuvent être découvertes en ne faisant qu'un minimum d'hypothèses a priori sur les phénomènes considérés. Elle illustre la synergie possible entre des mécanismes d'apprentissage portant sur des niveaux de description linguistique distincts, et cherche à déterminer quand et pourquoi cette coopération échoue. Elle conclut que la tension entre l'universalité de la distinction consonnes-voyelles et la spécificité de la structuration racine-schème est cruciale pour expliquer les forces et les faiblesses d'une telle approche. ABSTRACT This dissertation is concerned with the development of algorithmic methods for the unsupervised learning of natural language morphology, using a symbolically transcribed wordlist. It focuses on the case of languages approaching the introflectional type, such as Arabic or Hebrew. The morphology of such languages is traditionally described in terms of discontinuous units: consonantal roots and vocalic patterns. Inferring this kind of structure is a challenging task for current unsupervised learning systems, which generally operate with continuous units. In this study, the problem of learning root-and-pattern morphology is divided into a phonological and a morphological subproblem. The phonological component of the analysis seeks to partition the symbols of a corpus (phonemes, letters) into two subsets that correspond well with the phonetic definition of consonants and vowels; building around this result, the morphological component attempts to establish the list of roots and patterns in the corpus, and to infer the rules that govern their combinations. We assess the extent to which this can be done on the basis of two hypotheses: (i) the distinction between consonants and vowels can be learned by observing their tendency to alternate in speech; (ii) roots and patterns can be identified as sequences of the previously discovered consonants and vowels respectively. The proposed algorithm uses a purely distributional method for partitioning symbols. Then it applies analogical principles to identify a preliminary set of reliable roots and patterns, and gradually enlarge it. This extension process is guided by an evaluation procedure based on the minimum description length principle, in line with the approach to morphological learning embodied in LINGUISTICA (Goldsmith, 2001). The algorithm is implemented as a computer program named ARABICA; it is evaluated with regard to its ability to account for the system of plural formation in a corpus of Arabic nouns. This thesis shows that complex linguistic structures can be discovered without recourse to a rich set of a priori hypotheses about the phenomena under consideration. It illustrates the possible synergy between learning mechanisms operating at distinct levels of linguistic description, and attempts to determine where and why such a cooperation fails. It concludes that the tension between the universality of the consonant-vowel distinction and the specificity of root-and-pattern structure is crucial for understanding the advantages and weaknesses of this approach.

Design and Performance Verifcation of Ultra-High Performance Concrete Piles for Deep Foundations Final Report, November 2008

The strategic plan for bridge engineering issued by AASHTO in 2005 identified extending the service life and optimizing structural systems of bridges in the United States as two grand challenges in bridge engineering, with the objective of producing safer bridges that have a minimum service life of 75 years and reduced maintenance cost. Material deterioration was identified as one of the primary challenges to achieving the objective of extended life. In substructural applications (e.g., deep foundations), construction materials such as timber, steel, and concrete are subjected to deterioration due to environmental impacts. Using innovative and new materials for foundation applications makes the AASHTO objective of 75 years service life achievable. Ultra High Performance Concrete (UHPC) with compressive strength of 180 MPa (26,000 psi) and excellent durability has been used in superstructure applications but not in geotechnical and foundation applications. This study explores the use of precast, prestressed UHPC piles in future foundations of bridges and other structures. An H-shaped UHPC section, which is 10-in. (250-mm) deep with weight similar to that of an HP10×57 steel pile, was designed to improve constructability and reduce cost. In this project, instrumented UHPC piles were cast and laboratory and field tests were conducted. Laboratory tests were used to verify the moment-curvature response of UHPC pile section. In the field, two UHPC piles have been successfully driven in glacial till clay soil and load tested under vertical and lateral loads. This report provides a complete set of results for the field investigation conducted on UHPC H-shaped piles. Test results, durability, drivability, and other material advantages over normal concrete and steel indicate that UHPC piles are a viable alternative to achieve the goals of AASHTO strategic plan.

Development of LRFD Procedures for Bridge Pile Foundations in Iowa: Volume II: Field Testing of Steel Piles in Clay, Sand, and Mixed Soils and Data Analysis, September 2011

In response to the mandate on Load and Resistance Factor Design (LRFD) implementations by the Federal Highway Administration (FHWA) on all new bridge projects initiated after October 1, 2007, the Iowa Highway Research Board (IHRB) sponsored these research projects to develop regional LRFD recommendations. The LRFD development was performed using the Iowa Department of Transportation (DOT) Pile Load Test database (PILOT). To increase the data points for LRFD development, develop LRFD recommendations for dynamic methods, and validate the results ofLRFD calibration, 10 full-scale field tests on the most commonly used steel H-piles (e.g., HP 10 x 42) were conducted throughout Iowa. Detailed in situ soil investigations were carried out, push-in pressure cells were installed, and laboratory soil tests were performed. Pile responses during driving, at the end of driving (EOD), and at re-strikes were monitored using the Pile Driving Analyzer (PDA), following with the CAse Pile Wave Analysis Program (CAPWAP) analysis. The hammer blow counts were recorded for Wave Equation Analysis Program (WEAP) and dynamic formulas. Static load tests (SLTs) were performed and the pile capacities were determined based on the Davisson’s criteria. The extensive experimental research studies generated important data for analytical and computational investigations. The SLT measured loaddisplacements were compared with the simulated results obtained using a model of the TZPILE program and using the modified borehole shear test method. Two analytical pile setup quantification methods, in terms of soil properties, were developed and validated. A new calibration procedure was developed to incorporate pile setup into LRFD.

Les sociologues ont-ils perdu l'esprit ? À la recherche des structures élémentaires de la vie sociale

Ce texte est un « droit de réponse » par les auteurs de l'article « Vers un naturalisme social. À la croisée des sciences sociales et des sciences cognitives », publié par SociologieS en octobre 2011, au débat qu'il a suscité. Après une brève mise au point sur la forme même du débat, ainsi que sur les dissensions ponctuelles qui opposent les différents protagonistes, l'article répond aux inquiétudes parfaitement légitimes et aux questions de fond que soulève le naturalisme social.

Improved segmentation of deep brain grey matter structures using magnetization transfer (MT) parameter maps.

Basal ganglia and brain stem nuclei are involved in the pathophysiology of various neurological and neuropsychiatric disorders. Currently available structural T1-weighted (T1w) magnetic resonance images do not provide sufficient contrast for reliable automated segmentation of various subcortical grey matter structures. We use a novel, semi-quantitative magnetization transfer (MT) imaging protocol that overcomes limitations in T1w images, which are mainly due to their sensitivity to the high iron content in subcortical grey matter. We demonstrate improved automated segmentation of putamen, pallidum, pulvinar and substantia nigra using MT images. A comparison with segmentation of high-quality T1w images was performed in 49 healthy subjects. Our results show that MT maps are highly suitable for automated segmentation, and so for multi-subject morphometric studies with a focus on subcortical structures.

Generation of dynamic structures in nonequilibrium reactive membranes

We present a nonequlibrium approach for the study of a flexible bilayer whose two components induce distinct curvatures. In turn, the two components are interconverted by an externally promoted reaction. Phase separation of the two species in the surface results in the growth of domains characterized by different local composition and curvature modulations. This domain growth is limited by the effective mixing due to the interconversion reaction, leading to a finite characteristic domain size. In addition to these effects, first introduced in our earlier work [ Phys. Rev. E 71 051906 (2005)], the important new feature is the assumption that the reactive process actively affects the local curvature of the bilayer. Specifically, we suggest that a force energetically activated by external sources causes a modification of the shape of the membrane at the reaction site. Our results show the appearance of a rich and robust dynamical phenomenology that includes the generation of traveling and/or oscillatory patterns. Linear stability analysis, amplitude equations, and numerical simulations of the model kinetic equations confirm the occurrence of these spatiotemporal behaviors in nonequilibrium reactive bilayers.

An ultrastructural approach to analogies between fungal structures and needle fibre calcite

Needle fiber calcite (NFC) is an ubiquitous terrestrial secondary calcium carbonate mineral often associated with calcitic nanofibers. NFC's origin has been debated for a long time and a fungal origin is often proposed. Fungi are known to be involved in mineral weathering and production of metal oxalate, but little information exists regarding the genesis of other minerals, such as calcite. In this study, a comparison of similar ultrastructural characteristics of fungal hyphae and NFC has been performed to highlight analogies between both features. These analogies clearly demonstrate the probable close relationship between fungal filaments (hyphae and rhizomorphs) and NFC and its associated nanofibers.

Dynamics of transient domain structures in nematic liquid crystals: The splay Fréedericksz transition

We discuss the dynamics of the transient pattern formation process corresponding to the splay Fréedericksz transition. The emergence and subsequent evolution of the spatial periodicity is here described in terms of the temporal dependence of the wave numbers corresponding to the maxima of the structure factor. Situations of perpendicular as well as oblique field-induced stripes relative to the initial orientation of the director are both examined with explicit indications of the time scales needed for their appearance and posterior development.

Influence of bedrock structures on the spatial pattern of erosional landforms in small alpine catchments

Structural settings and lithological characteristics are traditionally assumed to influence the development of erosional landforms, such as gully networks and rock couloirs, in steep mountain rock basins. The structural control of erosion of two small alpine catchments of distinctive rock types is evaluated by comparing the correspondences between the orientations of their gullies and rock couloirs with (1) the sliding orientations of potential slope failures mechanisms, and (2) the orientation of the maximum joint frequency, this latter being considered as the direction exploited primarily by erosion and mass wasting processes. These characteristic orientations can be interpreted as structural weaknesses contributing to the initiation and propagation of erosion. The morphostructural analysis was performed using digital elevation models and field observations. The catchment comprised of magmatic intrusive rocks shows a clear structural control, mostly expressed through potential wedges failure. Such joint configurations have a particular geometry that encourages the development of gullies in hard rock, e.g. through enhanced gravitational and hydrological erosional processes. In the catchment underlain by sedimentary rocks, penetrative joints that act as structural weaknesses seem to be exploited by gullies and rock couloirs. However, the lithological setting and bedding configuration prominently control the development of erosional landforms, and influence not only the local pattern of geomorphic features, but the general morphology of the catchment. The orientations of the maximum joint frequency are clearly associated with the gully network, suggesting that its development is governed by anisotropy in rock strength. These two catchments are typical of bedrock-dominated basins prone to intense processes of debris supply. This study suggests a quantitative approach for describing the relationship between bedrock jointing and geomorphic features geometry. Incorporation of bedrock structure can be relevant when studying processes governing the transfer of clastic material, for the assessment of sediment yields and in landforms evolution models.

Cellular perspectives on the glutamate-monoamine interactions in limbic lobe structures and their relevance for some psychiatric disorders.

Dopaminergic, serotonergic and noradrenergic nuclei form the trimonoamine modulating system (TMMS). This system modulates emotional/motivational activities mediated by the limbic circuitry, where glutamate is the major excitatory neurotransmitter. Two main concepts are the basis of this review. First, since 1950 and the discovery of the antipsychotic activity of the dopamine D2 receptor antagonist chlorpromazine, it appears that drugs that can modulate the TMMS possess therapeutic psychiatric properties. Second, the concept of glutamate/trimonoamine imbalance in the cortico-striato-thalamo-cortical loop that has been so successful in explaining the pathophysiology of Parkinson disease has been applied in the pathophysiology of schizophrenia. This review will focus on the complex interactions between the fast synaptic glutamatergic transmission and the TMMS in specific parts of the limbic lobe and we will try to link these interactions to some psychiatric disorders, mainly depression, schizophrenia and drug addiction.