Literatura académica sobre el tema "Modèle en masses neurales"
Crea una cita precisa en los estilos APA, MLA, Chicago, Harvard y otros
Consulte las listas temáticas de artículos, libros, tesis, actas de conferencias y otras fuentes académicas sobre el tema "Modèle en masses neurales".
Junto a cada fuente en la lista de referencias hay un botón "Agregar a la bibliografía". Pulsa este botón, y generaremos automáticamente la referencia bibliográfica para la obra elegida en el estilo de cita que necesites: APA, MLA, Harvard, Vancouver, Chicago, etc.
También puede descargar el texto completo de la publicación académica en formato pdf y leer en línea su resumen siempre que esté disponible en los metadatos.
Artículos de revistas sobre el tema "Modèle en masses neurales"
1
Ballouhey, Quentin, Laurence Richard, Laurent Fourcade, Ines Ben Rhaiem, Jean-Michel Vallat, Franck Sturtz y Sylvie Bourthoumieu. "Étude des dérivés des crêtes neurales dans un modèle murin d’atrésie intestinale". Morphologie 101, n.º 335 (diciembre de 2017): 188. http://dx.doi.org/10.1016/j.morpho.2017.07.073.
Texto completo
2
Remaud, Sylvie y Barbara Demeneix. "Les hormones thyroïdiennes régulent le destin des cellules souches neurales". Biologie Aujourd'hui 213, n.º 1-2 (2019): 7–16. http://dx.doi.org/10.1051/jbio/2019007.
Texto completoResumen
Les hormones thyroïdiennes (HT) sont essentielles pour le bon fonctionnement du cerveau tout au long de la vie des vertébrés, dès les stades précoces du neuro-développement. Des études épidémiologiques ont montré l’importance des HT de la mère pendant les premiers mois du développement fœtal : une déficience précoce en HT maternelles entraîne à long terme des altérations du développement cognitif et du comportement social de l’enfant. L’apport des modèles animaux, non seulement les modèles mammifères mais également les modèles alternatifs (poisson zèbre, xénope, poulet), a permis de décrypter les mécanismes cellulaires et moléculaires gouvernés par les HT lors du développement cérébral. En particulier le modèle rongeur a contribué à montrer que les HT ont également un rôle crucial chez l’adulte, principalement au sein de deux niches neurogéniques majeures, la zone sous-ventriculaire et la zone sous-granulaire de l’hippocampe où elles régulent finement le destin des cellules souches neurales (CSN). Une question essentielle en biologie des cellules souches est de comprendre, comment les HT gouvernent le devenir des CSN vers un destin neural ou glial et ce, afin de contribuer au développement du cerveau et de maintenir ses fonctions tout au long de la vie adulte dans des conditions physiologiques et lors d’un dommage cérébral (maladies neurodégénératives, maladies démyélinisantes ou accident vasculaire cérébral). Notre revue fait le point sur les connaissances actuelles sur le rôle d’un signal endocrinien clé, les HT, lors du développement du cerveau et de la neurogenèse adulte, et principalement chez les mammifères, notamment l’Homme.
3
Falek, L., H. Teffahi y A. Djeradi. "Simulation d’un modèle de la source vocale et détermination des paramètres de commande". Canadian Journal of Physics 87, n.º 2 (febrero de 2009): 111–16. http://dx.doi.org/10.1139/p08-104.
Texto completoResumen
Nous présentons une étude paramétrique d’un modèle de la source vocale, connu comme modèle à deux masses. Ce modèle nous a permis de caractériser la fréquence fondamentale de la source en fonction de la pression des poumons et la tension des cordes vocales. Les résultats d’une telle étude contribuent à une meilleure connaissance de la réponse d’un modèle des cordes vocales désormais classique.
4
El It, Fatima, Laurence Faivre, Christel Thauvin-Robinet, Antonio Vitobello y Laurence Duplomb. "Des organoïdes cérébraux pour la compréhension et la thérapie des maladies génétiques rares avec troubles neurodéveloppementaux". médecine/sciences 40, n.º 8-9 (agosto de 2024): 643–52. http://dx.doi.org/10.1051/medsci/2024100.
Texto completoResumen
Les maladies génétiques associées à des troubles neurodéveloppementaux (TND) regroupent plusieurs maladies pour lesquelles peu de traitements sont proposés. L’impossibilité d’accéder à des échantillons de cerveaux humains pour des études ex vivo, et les divergences entre l’homme et les modèles animaux rendent nécessaires de nouvelles approches de recherche. L’organoïde cérébral, une structure en trois dimensions, auto-organisée, et générée à partir de cellules souches pluripotentes induites, permet de reproduire les étapes de développement du cerveau humain, de la prolifération des cellules souches neurales à leur différenciation en neurones, en oligodendrocytes, ou en astrocytes. L’intérêt de ce modèle est désormais prouvé pour la compréhension du développement cérébral et pour la recherche de traitements. Après une présentation des cellules souches pluripotentes induites et des organoïdes, nous exposerons comment cette technique est actuellement déployée, en particulier pour étudier les mécanismes physiopathologiques résultant de variations génétiques pathogènes de gènes candidats de TND.
5
Teffahi, H. y S. Kherouf. "Effets du couplage source–conduit vocal sur le modèle à deux masses". Canadian Journal of Physics 88, n.º 9 (septiembre de 2010): 657–62. http://dx.doi.org/10.1139/p10-047.
Texto completoResumen
Nous présentons une simulation de la source vocale, connue comme modèle à deux masses. Ce modèle simule le fonctionnement des cordes vocales humaines. Une étude qualitative des effets du couplage source–conduit sur la forme et le spectre de l’onde de débit de la glotte est entreprise. Cette simulation va permettre de mieux cerner le processus acoustique de production de la parole.
6
Sauvage, Stéphane, Nadine Locoge, Hervé Plaisance, Patrice Coddeville y Jean-Claude Galloo. "Identification et contribution des sources de HCNM en zone rurale". Pollution atmosphérique, NS 2 (1 de septiembre de 2010): 131–42. http://dx.doi.org/10.54563/pollution-atmospherique.7121.
Texto completoResumen
Les composés hydrocarbonés non méthaniques (HCNM) jouent un rôle prépondérant dans la chimie de l’atmosphère puisqu’ils interviennent en tant que précurseurs de la formation de polluants secondaires comme l’ozone ou les aérosols organiques secondaires. Disposant d’une base de données conséquente d’au moins cinq années de mesures systématiques de 46 HCNM sur trois sites de l’observatoire MERA (Mesure et Évaluation en zone Rurale de la pollution Atmosphérique à longue distance), le modèle sources-récepteur PMF (Positive Matrix Factorization) a été appliqué pour mettre en évidence des associations inter-espèces et identifier des profils de sources à l’origine des concentrations mesurées sur les sites récepteurs. L’application de ce type de modèle sur des sites ruraux, éloignés des sources et à des espèces réactives comme les HCNM a constitué une nouvelle approche. L’hypothèse de la conservation des masses sur lequel s’appuie le modèle est discutable compte tenu de la réactivité différente des espèces. C’est pourquoi la réactivité a été prise en compte par l’ajout d’une composante d’incertitude comme données d’entrée du modèle. Une étude de sensibilité du modèle a permis de consolider les résultats obtenus par rapport aux choix méthodologiques. Cinq profils de sources communs aux trois sites ont été identifiés : les profils « chauffage résidentiel », « échappement automobile », « évaporation de carburants », « biogénique » et « sources lointaines ». Les contributions relatives de ces sources aux teneurs en HCNM sont cohérentes avec les données d’inventaire d’émissions et avec les résultats d’autres travaux. Elles font apparaître une hiérarchie différente et logique pour les trois sites. L’analyse de ces contributions par classe de rétrotrajectoires de masses d’air met en évidence les parts d’influence locale et continentale où les profils de sources urbaines contribuent davantage alors que la part d’influence océanique est dominée par le profil « sources lointaines ».
7
Kistler, Max. "Réduction fonctionnelle et réduction logique". Philosophiques 27, n.º 1 (2 de octubre de 2002): 27–38. http://dx.doi.org/10.7202/004938ar.
Texto completoResumen
RÉSUMÉ Kim attribue aux émergentistes un modèle de « réduction logique » dans lequel la prédiction ou l’explication d’une occurrence de la propriété réduite ne requiert, outre des informations sur le niveau réducteur, que des principes logiques et mathématiques. Sur la base de cette interprétation, je conteste deux thèses de Kim. La première concerne la légitimité du modèle émergentiste de réduction. J’essaie de montrer, à l’aide de l’exemple de l’addition des masses, que l’adoption de la réduction logique rendrait irréductibles certaines propriétés qui sont clairement réductibles. La deuxième est la thèse selon laquelle la réduction fonctionnelle correspond aux exigences émergentistes sur la réduction. Telle que Kim la caractérise, la réduction fonctionnelle comporte, outre une définition fonctionnelle de la propriété à réduire, l’indication des propriétés réalisatrices. Or cette information, qui correspond à la découverte d’une loi de correspondance (locale), est empirique et non seulement logique.
8
Majdoub, R., J. Gallichand y J. Caron. "Modélisation du lessivage des bromures dans des cases lysimétriques par la méthode numérique des lignes". Revue des sciences de l'eau 14, n.º 4 (12 de abril de 2005): 465–88. http://dx.doi.org/10.7202/705428ar.
Texto completoResumen
La modélisation du transport des solutés dans un milieu non-saturé repose habituellement sur l'équation de dispersion-advection (EDA). Un modèle numérique (TSOL) a été développé en couplant l'EDA avec l'équation de Richards et en incluant le prélèvement de l'eau par les plantes. La résolution numérique a été effectuée par la méthode numérique des lignes (MNL) qui présente une grande simplicité de programmation et résulte en une très bonne précision numérique. La précision de TSOL a été testée avec les résultats d'un modèle d'éléments finis (HYDRUS), et avec des données expérimentales (profils de concentration et masses de bromures récupérés) collectées pendant 195 jours dans trois cases lysimétriques installées sur un sol non remanié cultivé en pommes de terre. La comparaison entre TSOL et HYDRUS montre que la solution de la MNL est similaire à celle des éléments finis. Toutefois, pour l'ensemble des cases et des profondeurs, les modèles ont montré une surestimation des valeurs de concentration avec un écart moyen entre les concentrations mesurées et simulées par TSOL variant de 22 à 112 mg/l. Pour les cases B et C, l'erreur moyenne de biais hebdomadaire entre TSOL et les masses de bromures récupérés, était d'environ 5 mg/semaine. Dans le cas de la case A, l'erreur moyenne de biais hebdomadaire de TSOL était de 39 mg/semaine.
9
Schaub, Jean-Frédéric. "La Crise Hispanique de 1640 Le modèle des « révolutions périphériques » en question (note critique)". Annales. Histoire, Sciences Sociales 49, n.º 1 (febrero de 1994): 219–39. http://dx.doi.org/10.3406/ahess.1994.279254.
Texto completoResumen
L'histoire politique de l'Ancien Régime se donne pour objet l'étude des modalités de production d'une sphère publique à l'échelle de l'ensemble de la société. Le laboratoire français est peut-être le plus trompeur. En effet, si l'on en juge par la plupart des histoires récentes du XVIIe siècle, l'image lavissienne de Louis XIV est toujours d'actualité et la faveur dont jouit l'œuvre de Tocqueville accuse le trait : l'absolutisme, à côté de la tyrannie symbolisée par la Révocation, c'est d'abord l'archéologie des masses de granit institutionnelles de la France.
10
Gallinari, F., S. Elmaleh y R. Ben Aïm. "Influence de la dissipation énergetique sur l'efficacité de la flottation à air dissous : analogie avec la floculation". Revue des sciences de l'eau 9, n.º 4 (12 de abril de 2005): 485–98. http://dx.doi.org/10.7202/705263ar.
Texto completoResumen
Les flottateurs à air dissous sont classiquement dimensionnés à partir de deux paramètres: le rapport mA/mS des masses d'air et de solides en présence et le temps de contact entre phases. Une série d'essais effectués sur cinq différentes unités continues ou discontinues montre que ces seules variables opératoires ne suffisent pas à déterminer l'efficacité. De plus, l'extrapolation des données obtenues sur un floculateur discontinu conduirait à des besoins en air dissous considérables pouvant limiter le développement industriel. La dissipation d'énergie, habituellement négligée, peut être quantifiée par le gradient de vitesse tel qu'il a été introduit en théorie de la floculation. De plus, une analogie entre la capture bulles-particules et le processus de floculation des particules primaires sur les flocs déjà formés permet d'étendre les équations de vitesse de la floculation et d'obtenir un modèle cinétique où interviennent seulement le gradient de vitesse et la concentration de particules; ce modèle remplace avantageusement l'approche classique qui considère la flottation comme un processus du premier ordre par rapport aux particules. L'efficacité d'une cellule discontinue ou d'un floculateur piston est alors fonction du seul nombre de Camp. Les résultats montrent l'existence d'un intervalle optimal pour le gradient de vitesse, 3000 à 4000 s-¹, et pour le nombre de Camp 105 à 106. Le modèle devra être amélioré par introduction de la tension critique de mouillage des particules.
Tesis sobre el tema "Modèle en masses neurales"
1
Kuchenbuch, Mathieu. "Modélisation computationelle de l'épilepsie avec crises focales migrantes du nourrisson". Thesis, Rennes 1, 2019. http://www.theses.fr/2019REN1B062.
Texto completoResumen
L’épilepsie avec crises focales migrantes du nourrisson se caractérise par l’apparition de crises focales avant 6 mois qui s’intensifient jusqu’à une phase d’orage où il y a des crises migrantes. Les mutations gain-defonction du gène KCNT1 sont les principales mutations responsables de cette épilepsie. Nous nous sommes intéressés à une cohorte de patients ayant une mutation du gène KCNT1 et cette épilepsie afin de mieux comprendre ce syndrome en vue de le modéliser. D’abord, nous avons précisé la clinique de ces patients, notamment le mauvais devenir à long terme, la forte mortalité, la microcéphalie et la présence de symptômes extra-neurologiques. Puis, nous avons déterminé, via l’étude des EEG ictaux, que les crises migrantes n’étaient pas chaotiques mais correspondaient plutôt à un type de propagation et nous avons identifié des marqueurs spécifiques de cette épilepsie. Ensuite, nous avons montré que la majorité des mutations KCNT1 semblaient se regrouper dans des "points chauds" et qu’il n’y avait pas de corrélation génotype-phénotype stricte. Finalement, nous avons modélisé cette épilepsie à l’échelle microscopique et mésoscopique. Les résultats préliminaires montraient une baisse de l’excitation, une chute de l’inhibition et l’implication du GABA dépolarisant. Nous discutons ensuite des différents aspects de nos travaux à la lumière de la littérature et décrivons les perspectives ouvertes par cette thèse d’un point de vue fondamental, clinique et physiologiqueEpilepsy in infancy with migrating focal seizures is characterized by focal seizures beginning before 6 months that intensify to a stormy phase where so-called migrating focal seizures appear. The gain-of-function mutations of the KCNT1 gene are the main causes of this epilepsy. We focused on a cohort of patients with a KCNT1 mutations and this epilepsy to better understand this syndrome in order to model it. First, we specified the clinic for these patients, including long-term poor outcomes, high mortality, microcephaly and the presence of extra-neurological symptoms. Then, we determined, through the study of ictal EEGs, that migrating seizures were not chaotic but rather corresponded to a type of propagation and we have identified specific markers of this epilepsy. Then, we showed that the majority of KCNT1 mutations appeared to cluster in "hot spots" and that there was no strict genotypephenotype correlation. Finally, we modelled this epilepsy at microscopic and mesoscopic levels. Preliminary results showed a decrease in excitation, a fall in inhibition and involvement of depolarizing GABA. We then discuss the different aspects of our work in the light of the literature and describe the perspectives opened by this thesis from a fundamental, clinical and physiological point of views
2
Condy, Carine. "La distractibilité : bases neurales, pharmacologie et modèles expérimentaux". Paris 6, 2007. https://tel.archives-ouvertes.fr/tel-00808982.
Texto completoResumen
La distractibilité est un syndrome qui se caractérise par une réactivité exagérée d’un individu vis-à-vis de son environnement. Il a été démontré que des atteintes limitées au cortex préfrontal dorsolatéral (CPFDL) sont responsables d’un tel syndrome chez l’homme. Ce trouble peut être quantifié par un test spécifique et sensible appelé « test des antisaccades ». Ce syndrome se rencontre dans de nombreuses pathologies neurodégénératives telles que la Paralysie supranucléaire progressive (PSP) ou encore psychiatriques telles que la schizophrénie. L’objectif de cette recherche est de tenter de mieux comprendre ce syndrome de distractibilité à la fois sur le plan anatomique et pharmacologique. Le projet anatomique débute par une étude chez l’homme de lésions au niveau sous cortical pour déterminer l’implication réelle du sous cortex dans ce syndrome. Afin d’apporter des données indiscutables sur le rôle du CPFDL dans la distractibilité, nous avons réalisé une étude chez le primate de microinjections intracérébrales de muscimol au niveau de cette aire cérébrale. L’approche pharmacologique a consisté à réaliser un modèle animal de distractibilité en reproduisant chez le primate ce syndrome par administration de kétamine à dose subanesthésique. Celle-ci est connue pour induire des symptômes « shizophréniques like » avec des troubles liés à un dysfonctionnement du cortex préfrontal. La seconde étude a pour but d’évaluer l’effet d’un inhibiteur de l’acétylcholine estérase sur le syndrome de distractibilité chez des patients atteints de PSP. Cette recherche, en particulier les études chez le primate devraient permettre à terme de développer une thérapeutique pour enrayer ce trouble.
3
Condy, Carine. "LA DISTRACTIBILITE : BASES NEURALES, PHARMACOLOGIE ET MODELES EXPERIMENTAUX". Phd thesis, Université Pierre et Marie Curie - Paris VI, 2007. http://tel.archives-ouvertes.fr/tel-00808982.
Texto completoResumen
La distractibilité est un syndrome qui se caractérise par une réactivité exagérée d'un individu vis-à-vis de son environnement. Il a été démontré que des atteintes limitées au cortex préfrontal dorsolatéral (CPFDL) sont responsables d'un tel syndrome chez l'homme. Ce trouble peut être quantifié par un test spécifique et sensible appelé " test des antisaccades ". Ce syndrome se rencontre dans de nombreuses pathologies neurodégénératives telles que la Paralysie supranucléaire progressive (PSP) ou encore psychiatriques telles que la schizophrénie. L'objectif de cette recherche est de tenter de mieux comprendre ce syndrome de distractibilité à la fois sur le plan anatomique et pharmacologique. Le projet anatomique débute par une étude chez l'homme de lésions au niveau sous cortical pour déterminer l'implication réelle du sous cortex dans ce syndrome. Afin d'apporter des données indiscutables sur le rôle du CPFDL dans la distractibilité, nous avons réalisé une étude chez le primate de microinjections intracérébrales de muscimol au niveau de cette aire cérébrale. L'approche pharmacologique a consisté à réaliser un modèle animal de distractibilité en reproduisant chez le primate ce syndrome par administration de kétamine à dose subanesthésique. Celle-ci est connue pour induire des symptômes " shizophréniques like " avec des troubles liés à un dysfonctionnement du cortex préfrontal. La seconde étude a pour but d'évaluer l'effet d'un inhibiteur de l'acétylcholine estérase sur le syndrome de distractibilité chez des patients atteints de PSP. Cette recherche, en particulier les études chez le primate devraient permettre à terme de développer une thérapeutique pour enrayer ce trouble.
4
Nugroho, Dwiyoga. "La marée dans un modèle de circulation générale dans les mers indonésiennes". Thesis, Toulouse 3, 2017. http://www.theses.fr/2017TOU30089/document.
Texto completoResumen
Les mers Indonésiennes sont le siège de très fort courants de marée qui interagissent avec la topographie pour créer des ondes internes à la fréquence de la marée que l'on appelle marée interne. Certaines d'entres elles, vont se propager et se dissiper dans l'océan intérieur. Le mélange associé provoque la remontée d'eau plus froide et plus riche en nutriments en surface qui influence le climat tropical et toute la chaine des écosystèmes marins. Surveiller les ressources marines est l'objectif du projet INDESO, dont cette thèse fait partie. Prendre en compte le mélange induit par la marée interne n'est pas facile. En effet, le résoudre entièrement n'est pas possible car les échelles concernées par les différents processus des ondes internes varient de plusieurs milliers de kilomètres (propagation) à quelques centimètres/millimètres (dissipation). De plus en plus de scientifiques introduisent le forçage de la marée dans leur modèle mais sans savoir où va l'énergie et comment les ondes sont dissipées. Dans cette thèse nous cherchons à proposer des outils et des débuts de réponses pour participer à cette meilleure compréhension de la dissipation des ondes internes dans le modèle numérique d'océan NEMO. Nous proposons certaines quantifications que nous comparons aux anciennes paramétrisations. J'ai, tout d'abord, contribué à une étude d'INDESO sur la validation de NEMO grâce à de nombreux jeu de données. Ensuite, j'ai cherché à quantifier et à qualifier le mélange induit par l'introduction de la marée explicite dans le modèle, ainsi que son impact sur les masses d'eau. (c'est redit plus loin)Il produit un refroidissement de surface de 0.3°C avec des maxima atteignant 0.8°C au niveau des sites de génération des ondes internes. Le modèle reproduit 75% de l'énergie attendue de génération des ondes internes, en bon accord avec des études précédentes. L'essentiel de la dissipation a lieu horizontalement (19GW) est proche de celle induite par la paramétrisation couramment utilisée (16GW), alors que, dans la réalité, on s'attend principalement à une dissipation réalisée grâce à des processus verticaux. Le modèle, au dessus des zones de génération, est de façon surprenante en très bon accord avec les mesures in situ de dissipation obtenues lors de la campagne INDOMIX. Par contre, dans les régions distantes des sources de génération, le modèle surestime le mélange par rapport aux observations d'INDOMIX. Dans la dernière partie de cette thèse j'ai commencé à apporter des éléments de réponse à la quantification des puits d'énergie dans NEMO. J'ai pour cela travaillé avec le cas test COMODO, qui est une section d'un fluide stratifié constituée d'une plaine abyssale, d'un talus et d'un plateau, forcée par la marée et sans friction de fond. Le modèle T-UGOm, un modèle hydrodynamique de marée, est comparé au modèle NEMO. Dans ce cadre, nous avons développé une méthode originale pour séparer la marée barotrope de la marée barocline. Elle repose sur la projection en modes normaux. Cette méthode donne, à première vue, des résultats similaires à ceux obtenus grâce à la méthode plus classique de soustraction par la moyenne verticale. Cependant, lorsque l'on regarde plus en détail les diagnostiques d'énergie on trouve que la méthode de projection en modes normaux offre une plus grande précision et un plus grand réalisme pour séparer la marée barotrope de la marée barocline. Plus on monte dans des modes élevés plus les longueurs ondes se raccourcissent dans NEMO par rapport à T-UGOm. Par ailleurs, NEMO dissipe la marée barotrope dans la plaine abyssale, alors qu'il n'y a explicitement pas de friction. Ce ne peut pas être la diffusion verticael ou horizontale qui est à l'œuvre ici, car il n'y a pas de raison physique pour une diffusion sur un fond plat. Le meilleur candidat pour expliquer cette diffusion serait le couplage 2D/3D du time splitting de NEMO. Un travail est en cours pour appliquer cette méthode sur l'ensemble de l'archipel IndonésienIn the Indonesian seas, large tidal currents interact with the rough topography and create strong internal waves at the tidal frequency, called internal tides. Part of them will eventually propagate and dissipate far away from generation sites. Their associated mixing upwells cold and nutrient-rich water that prove to be critical for climate system and for marine resources. This thesis uses the physical ocean general circulation model, NEMO, as part of the INDESO project that aims at monitoring the Indonesian marine living resources. Models not taking into account tidal missing are unable to correctly reproduce the vertical structure of watermasses in Indonesian seas. However, taking into account this mixing is no simple task as the phenomena involved in tidal mixing cover a wide spectrum of spatial scales. Internal tides indeed propagate over thousands of kilometres while dissipation and mixing occurs at centimetric to millimetric scales. A model capable of resolving all these processes at the same time does not exist. Until now scientists either parameterised the tidal mixing or used models which only partly resolve internal tides. More and more scientists introduce explicit tidal forcing in their models but without knowing where the energy is going and how the internal tides are dissipated. This thesis intends to quantify energy dissipation in NEMO forced with explicit tidal forcing and compares it to the dissipation induced by the currently used parameterization. This thesis also provides new results about the quantification of the tidal energy budget in NEMO. I first contributed to an INDESO study that aimed at validating the model against several observation data sets. In a second and third study, I investigated the mixing produced in the model by explicit tidal forcing and its impact on water mass. Explicit tides forcing proves to produce a mixing comparable to the one produced by the parameterization. It also produces a significant cooling of 0.3 °C with maxima reaching 0.8°C in the areas of internal tide generation. The cooling is stronger on austral winter. The spring tides and neap tides modulate this impact by 0.1°C to 0.3°C. The model generates 75% of the expected internal tides energy, in good agreement with other previous studies. In the ocean interior, most of it is dissipated by horizontal momentum dissipation (19 GW), while in reality one would expect dissipation through vertical possesses. This value is close to the dissipation induced by the parameterization (16 GW). The mixing is strong over generation sites, and only 20% remains for far field dissipation mainly in the Banda and Sulawesi Seas. The model and the recent INDOMIX cruise [Koch-Larrouy et al. (2015)], which provided direct estimates of the mixing, are surprisingly in good agreement mainly above straits. However, in regions far away from the energy generation sites where INDOMIX found NO evidence of intensified mixing, the model produces too strong mixing. The bias comes from the lack of specific set up of internal tides in the model. More work is thus needed to improve the modeled dissipation, which is a theme of active research for the scientific community. I dedicated the last part of my thesis to the quantification of tidal energy sinks in NEMO. I first worked on a simple academic case: the COMODO internal tides test case, which analyses the behaviour of a vertically stratified fluid forced by a barotropic flow interacting over an idealized abyssal plain/slope/shelf topography without bottom friction. The results of the finite element T-UGOm hydrodynamic model are compared with those of NEMO. The central issue in calculating tidal energy budget is the separation of barotropic and baroclinic precesses
5
Koch-Larrouy, Ariane. "Transformation des masses d'eau dans les mers indonésiennes". Paris 6, 2007. http://www.theses.fr/2007PA066621.
Texto completoResumen
L’objectif de cette thèse est de déterminer, de représenter et de quantifier les processus responsables de la très forte transformation des masses d’eau dans les mers indonésiennes. L’approche consiste à utiliser un modèle de circulation générale océanique forcé, en constante comparaison avec les observations. La première étape de ce travail de construire une nouvelle paramétrisation fondée sur des considérations physiques et contrainte par l’énergie de la marée interne. Elle permet de reproduire les masses d’eau dans chacun des sous-bassins des mers indonésiennes en bon accord avec les observations. Elle génère un mélange intense dans la thermocline dont l’intensité moyenne est très comparable à celle estimée à partir des observations. Cette simulation et l’utilisation conjointe de diagnostiques complémentaires très puissants ont permis de quantifier la transformation des masses d’eau. Cette solution a été confrontée aux observations afin de fournir de nouvelles explications permettant d’interpréter les observations. Il est montré que la marée joue un rôle dominant dans cette transformation. Elle est responsable de flux diapycnaux très intenses seuls capables de reproduire la langue d’eau froide et douce dans la thermocline, caractéristique de l’eau indonésienne dans l’Océan Indien. Ce mélange intense dans la thermocline rétroagit sur le forçage et permet à l’océan d’absorber deux fois plus de chaleur (20 W/m2). Finalement, la transformation est étudiée le long des trajectoires de l’ITF et les eaux formées sont reliées à leur origine avec une quantification des échanges de sel et de température.
6
Kalantari, Saman. "Introduction de fonctionnalités de changements d'états topologiques dans le formalisme de modélisation et de simulation CORDIS-ANIMA". Thesis, Grenoble, 2014. http://www.theses.fr/2014GRENS004.
Texto completoResumen
Dans cette recherche, nous étudions les problèmes de modélisation de discontinuités topologiques d'objets physiques déformables. Parmi toutes les discontinuités topologiques possibles, nous nous intéressons à celles que nous avons qualifiées de « discontinuités de 1 vers N » dans le contexte de la modélisation physique. Entrent dans cette catégorie les phénomènes de fractures, de cassures, de déchirures et de fissures. De nombreux travaux se sont intéressés à la modélisation et la simulation de ce type de phénomènes. Plusieurs méthodes ont été proposées, à partir de systèmes de modélisation différents parmi lesquels les plus courants sont : modélisation masses – ressorts sur maillages, méthodes d'éléments finis avec et sans maillage. Une question commune qui se pose dans toutes les méthodes discutées dans cette thèse est la question des endroits fonctionnels où apparaissent et se propagent les discontinuités topologiques. Nous étudions ce problème à travers plusieurs approches sur la modélisation physique de discontinuités topologiques. Nous proposons ensuite notre approche, qui se fonde sur le formalisme masses – interactions. Nous montrons en quoi ce type de modélisation offre certaines propriétés natives pour traiter cette question. En particulier, le fait qu'elles soient par principe affranchies de toute notion de contigüité spatiale et de continuité matérielle leur permet de poser de manière très générale la question des changements d'état physique à l'origine des discontinuités topologiques observées. Nous montrons qu'une très grande majorité de changements d'états physiques peuvent se modéliser par des variations des paramètres physiques d'interactions physiques non linéaires en lieu et place de changements structurels adoptés dans majorité des travaux menés par des méthodes courante et qui conduisent à des modifications très importantes et complexes du modèle physique. Néanmoins, cela induit, des effets d'enlèvements de matière aux endroits du changement d'état puisque celui –ci est porté par l'interaction, c'est à dire une contrainte physique qui est placée entre les points matériels. Pour s'affranchir de ces effets, nous proposons alors une nouvelle méthode, qui consiste à modéliser les discontinuités topologiques sur un élément de masse physique et non pas sur un élément de contrainte physique. Nous appelons ce dispositif « MAT éclatable ». Le processus de modélisation avec les MAT éclatables s'inscrit dans le processus de modélisation masses-interactions tel que défini dans le formalisme de modélisation et de simulation de CORDIS – ANIMA. La méthode proposée permet d'effectuer des changements topologiques par modèle physique par changements structurels, mais différemment de ceux plus couramment effectuées dans les méthodes usuelles (tels que par exemple, ajout ou suppression de ressorts ou de masses), la méthode des MAT éclatables permet de maîtriser le fonctionnement du modèle physique en termes d'optimalité des transformations, de stabilité du temps de calcul et de stabilité numérique. La méthode des MAT éclatables a été développée sous forme d'un dispositif de modélisation, destiné aux utilisateurs de notre outil de modélisation CORDIS-ANIMA. Pour illustrer et valider la méthode, nous proposons un ensemble de modèles physiques modélisation des fractures et déchirures multiples apparaissant et se propageant dans des objets très déformables, des effilochures, ou des fragmentationsL'auteur n'a pas fourni de résumé en anglais
7
Pla, Patrick. "Les cellules embryonnaires souches sauvages et mutantes : un nouveau modèle de la migration et de la différenciation de crêtes neurales". Paris 11, 2003. http://www.theses.fr/2003PA112124.
Texto completoResumen
Lors du développement embryonnaire des Vertébrés, les cellules de crête neurale (NCC) troncales qui migrent sur la voie dorso-ventrale donnent naissance aux neurones et aux cellules gliales du système nerveux périphérique. Les NCC qui migrent sur la voie dorso-latérale sont les mélanoblastes et donnent naissance aux mélanocytes. Pour mieux connaitre les mécanismes moléculaires déterminant l'orientation vers la voie de migration dorso-latérale, nous avons développé un modèle cellulaire qui utilise les cellules embryonnaires souches (ES) de souris, des cellules totipotentes issues de la masse cellulaire interne des blastocystes. Ces cellules sont capables de migrer sur les 2 voies de migration des NCC, une fois greffées dans l'embryon de poulet au niveau de " l'aiguillage " entre ces deux voies. Grâce à ce modèle, nous avons montré que l'intégrine β1, une molécule liant la matrice extracellulaire au cytosquelette, et le récepteur aviaire b2 aux endothélines (Ednrb2) ont un rôle déterminant dans la migration dorso-latérale. Un récepteur murin proche de Ednrb2, Ednrb, n'a pas le même effet. Ednrb2 est capable de stimuler l'activité de FAK et de paxilline, des tyrosine-kinases impliquées dans la migration et la signalisation des intégrines. De plus, il est possible de favoriser in vitro la détermination des cellules ES vers le lignage mélanocytaire. Des cellules ES exprimant un gène rapporteur sous le contrôle d'un promoteur mélanoblaste-spécifique, Dct, ont été établies et le déroulement du processus de détermination/différenciation in vitro des cellules ES en mélanocytes a été suivi. Les cellules ES exprimant Ednrb2 ou Ednrb s'orientent plus facilement vers le lignage mélanocytaire. L'addition du ligand endothéline stimule également la prolifération des mélanoblastes dérivés des cellules ES. Les travaux de cette thèse ont ainsi permis la mise au point, le perfectionnement et l'application d'un nouveau modèle d'étude du développement des NCCTruncal neural crest cells (NCC) that migrate along the dorsoventral pathway give rise to neurons and glial cells of the peripheral nervous system. NCC that migrate along the dorsolateral pathway are called melanoblasts and give rise to melanocytes. To better determine the molecular mechanisms that determine the orientation towards the dorsolateral pathway, we developped a cellular model using mouse embryonic stem cells (ES cells). These cells are able to migrate on both NCC pathways, when they are grafted in the chicken embryo. Using this model, we found that integrin beta1, a protein linking the extracellular matrix with the cytoskeleton, and Ednrb2 have a crucial role in the dorsolateral migration. Ednrb2 is able to stimulate the activity of FAK and paxillin, two tyrosine kinases involved in migration. Moreover, we differentiated ES cells into melanocytes in vitro. ES expressing a reporter gene under the control of a melanoblast-specific promoter (Dct : :lacZ) were established and the cellular events such as determination, proliferation and differentiation were observed during their differentiation in vitro. ES cells expressing Ednrb2 are more easily recruited to the melanoçyte lineage. The addition of the ligand endothelin stimulated also the proliferation of ES-cell derived melanoblasts. Altogether, this work show the development and the application of novel model of NCC development
8
Dugault, Edouard. "Etude du transport et de l'évolution physico-chimique de masses d'air européennes au-dessus de l'Atlantique Nord sous des conditions anticycloniques". Paris 6, 2002. http://www.theses.fr/2002PA066518.
Texto completo
9
Vallet, Anais. "Etude de la balance Excitatiοn/Ιnhibitiοn de régiοns cérébrales impliquées dans une tâche de cοntrôle inhibiteur : mοdélisatiοn de dοnnées οbtenues en Ιmagerie par Résοnance Μagnétique fοnctiοnnelle et inversiοn". Electronic Thesis or Diss., Normandie, 2024. http://www.theses.fr/2024NORMC014.
Texto completoResumen
En psychologie, le contrôle inhibiteur est un mécanisme cognitif qui permet de stopper une ré-ponse motrice, émotionnelle ou cognitive non adaptée pour la réalisation d’un but désiré. Au niveaucérébral, le contrôle inhibiteur est associé au fonctionnement en réseau de régions cérébrales, quipeut être mesuré à partir du signal BOLD en IRMf. Des régions de contrôle préfrontales abaissentl’activité BOLD de régions cibles. L’IRMf permet de mesurer de manière indirecte l’activité desneurones. Comment peut-on alors inférer à partir de données d’IRMf des propriétés excitatriceset inhibitrices (E/I) neurales au sein de régions cérébrales impliquées dans une tâche de contrôleinhibiteur ?Nous partons d’un modèle biophysique non linéaire, hiérarchique qui décrit les évolutions tempo-relles des activités neurales excitatrice et inhibitrice par région (Naskar et al., 2021). Ces varia-tions d’activité produisent des changements BOLD dans chaque région cérébrale. L’analyse de cemodèle nous permet de : 1) identifier des paramètres neuraux de la balance E/I ; 2) montrer quel’augmentation d’activité BOLD d’une région de contrôle ne permet pas d’abaisser l’activité BOLDd’une région cible parce que les régions sont connectées par leurs neurones excitateurs uniquement ;3) proposer une nouvelle architecture de connectivité pour le permettre ; 4) étudier comment labaisse d’activité de la région cible dépend de la balance E/I dans la cible. Nous proposons alorsune nouvelle procédure d’inversion. Nous en vérifions la fiabilité avec des simulations, avant deprésenter une preuve de concept sur les données d’un sujet pendant une tâche de Think/NoThink,un paradigme d’étude du contrôle inhibiteur des intrusions mnésiques (Mary et al., 2020)In psychology, inhibitory control is a cognitive mechanism that stops a motor, emotional orcognitive response from achieving a desired goal. At cerebral level, inhibitory control is associatedwith a network of brain regions, whose function may be measured using BOLD signals from fMRI.Prefrontal control regions lower the BOLD activity of target regions. fMRI provides an indirectmeasure of the activity of neurons. How can we then infer from fMRI data, neural excitatory andinhibitory (E/I) properties of brain regions involved in an inhibitory control task ?We start with a non-linear biophysical model that describes by region the temporal evolutionof neural excitatory and inhibitory activities (Naskar et al., 2021). These variations in activityproduce BOLD changes in each brain region. Analysis of this model enables us to : 1) identifyneural parameters of the E/I balance ; 2) show that increasing the BOLD activity of a controlregion does not lower the BOLD activity of a target region, since these regions are connected bytheir excitatory neurons only ; 3) propose a new connectivity architecture to enable this ; 4) studyhow the lowering of activity in the target region depends on the E/I balance in the target region.We then propose a new inversion procedure. We check its reliability through simulations, beforepresenting a proof-of-concept using real data from a subject during a Think/No-Think task, aparadigm used for studying the inhibitory control of memory intrusions (Mary et al., 2020)
10
Maire, Cecile. "Fonction des facteurs de transcription Olig1 et Olig2 dans les cellules souches neurales du système nerveux central : Etude d'un modèle de souris transgéniques d'expression inductible". Paris 5, 2007. http://www.theses.fr/2007PA05D030.
Texto completoResumen
L'objectif de ma thèse a été de définir l'impact de la sur-expression des facteurs de transcription Olig sur la différenciation des cellules souches neurales en oligodendrocyte dans un modèle de souris transgénique. L'expression des transgènes est induite par la doxycycline dans les cellules souches nestine+. Au cours du développement, l'expression forcée de Olig1 ou Olig2 induit une genèse ectopique d'oligodendrocytes. De plus, la sur-expression de Olig2 bloque la spécification des interneurones V3. En post-natal, la sur-expression de Olig2 dans les zones germinatives provoque une myélinisation précoce et une augmentation du nombre d'astrocytes dans le corps calleux. L'ensemble de mes résultats indique que la sur-expression des facteurs Olig pourrait constituer une stratégie thérapeutique intéressante pour promouvoir la régénération de la myéline dans des pathologies démyélinisantes, telle que la scérose en plaquesOlig1 and Olig2 are b-HLH transcription factors involved in oligodendrocyte development in the central nervous system. My project aims to analyse the effect of Olig gene over-expression in neural stem cells. Therefore, I designed and analyzed transgenic mice models with inducible expression of Olig genes in nestin+ neural stem/progenitor cells (Tet-On system). At embryonic stages, forced expression of Olig1 and Olig2 leads to ectopic expression of oligodendrocyte markers. Moreover, Olig2 over-expression decreased V3 interneuron specification. At postnatal stage Olig2 over-expression in germinative area induced earlier myelination and astrocyte specification in corpus callosum. These transgenic mice provide a useful model to test whether forced expression of Olig genes represents a possible strategy to enhance remyelination in demyelinating disease such as multiple sclerosis
Libros sobre el tema "Modèle en masses neurales"
1
Butyrskiy, Evgeniy y Alexandr Matveev. Mathematical modeling of systems and processes. Strategy of the Future, 2022. http://dx.doi.org/10.37468/book_011222.
Texto completoResumen
The monograph considers the fundamentals of systems theory and mathematical modeling using the principles of the systems approach. In the monograph, much attention is paid to set-theoretic, dynamic, as well as aggregative and combined models. Based on the group-theoretical approach, a generalization of the theory of signals, their characteristics are considered, their classification and some theorems are carried out. A separate chapter is devoted to mathematical models of signal interaction with the propagation medium. The monograph also considers mathematical models of stochastic signal processing and control systems based on spline filtering, artificial intelligence models and neural networks. The monograph can be useful for a wide range of specialists in various fields of knowledge dealing with mathematical modeling in their research, and can also be used as a teaching aid for conducting both classroom and independent theoretical and practical classes with bachelors, masters, graduate students in the discipline "System Theory and System Analysis", "Mathematical Modeling" and "Optimal and Suboptimal Estimation of Random Processes and Systems".
2
Wendling, Fabrice y Fernando H. Lopes da Silva. Dynamics of EEGs as Signals of Neuronal Populations. Editado por Donald L. Schomer y Fernando H. Lopes da Silva. Oxford University Press, 2017. http://dx.doi.org/10.1093/med/9780190228484.003.0003.
Texto completoResumen
This chapter gives an overview of approaches used to understand the generation of electroencephalographic (EEG) signals using computational models. The basic concept is that appropriate modeling of neuronal networks, based on relevant anatomical and physiological data, allows researchers to test hypotheses about the nature of EEG signals. Here these models are considered at different levels of complexity. The first level is based on single cell biophysical properties anchored in classic Hodgkin-Huxley theory. The second level emphasizes on detailed neuronal networks and their role in generating different kinds of EEG oscillations. At the third level are models derived from the Wilson-Cowan approach, which constitutes the backbone of neural mass models. Another part of the chapter is dedicated to models of epileptiform activities. Finally, the themes of nonlinear dynamic systems and topological models in EEG generation are discussed.
Capítulos de libros sobre el tema "Modèle en masses neurales"
1
Holzinger, Andreas, Anna Saranti, Anne-Christin Hauschild, Jacqueline Beinecke, Dominik Heider, Richard Roettger, Heimo Mueller, Jan Baumbach y Bastian Pfeifer. "Human-in-the-Loop Integration with Domain-Knowledge Graphs for Explainable Federated Deep Learning". En Lecture Notes in Computer Science, 45–64. Cham: Springer Nature Switzerland, 2023. http://dx.doi.org/10.1007/978-3-031-40837-3_4.
Texto completoResumen
AbstractWe explore the integration of domain knowledge graphs into Deep Learning for improved interpretability and explainability using Graph Neural Networks (GNNs). Specifically, a protein-protein interaction (PPI) network is masked over a deep neural network for classification, with patient-specific multi-modal genomic features enriched into the PPI graph’s nodes. Subnetworks that are relevant to the classification (referred to as “disease subnetworks”) are detected using explainable AI. Federated learning is enabled by dividing the knowledge graph into relevant subnetworks, constructing an ensemble classifier, and allowing domain experts to analyze and manipulate detected subnetworks using a developed user interface. Furthermore, the human-in-the-loop principle can be applied with the incorporation of experts, interacting through a sophisticated User Interface (UI) driven by Explainable Artificial Intelligence (xAI) methods, changing the datasets to create counterfactual explanations. The adapted datasets could influence the local model’s characteristics and thereby create a federated version that distils their diverse knowledge in a centralized scenario. This work demonstrates the feasibility of the presented strategies, which were originally envisaged in 2021 and most of it has now been materialized into actionable items. In this paper, we report on some lessons learned during this project.
2
"Neuron Models and Neural Masses". En Neural Nets and Chaotic Carriers, 131–51. IMPERIAL COLLEGE PRESS, 2010. http://dx.doi.org/10.1142/9781848165915_0011.
Texto completo
3
Oshkhunov, Muaed M., Sergey I. Dosko y Aleksey Kh Tlibekov. "Solving Problems of Deformable Solid Mechanics by the Method of Dynamic Particles". En Advances in Transdisciplinary Engineering. IOS Press, 2024. http://dx.doi.org/10.3233/atde240665.
Texto completoResumen
A general method for solving all types of problems of “Solid mechanics” is given. A unified algorithm is proposed for solving the main types of problems of deformable solid mechanics: classical theory of elasticity, physically and geometrically nonlinear problems, dynamic problems of elasticity theory. A model of heat transfer in media with finite velocities is also proposed. The essence of the approach is to replace a continuous deformable medium with a system of interacting particles equivalent in physical and mechanical properties. The interaction potential between neighboring particles for an elastic isotropic continuous medium is determined by the Young’s modulus, Poisson’s ratio, or their analogues for heat transfer problems. The problems that must be solved for the effective use of a new method for solving problems of solid mechanics are outlined. Such a model is similar to dynamic models with concentrated masses, represented by systems of linear (or nonlinear) differential equations, algorithms and numerical methods for solving which are well known. However, it is necessary to ensure the accuracy of numerical integration and reduce the number of computational operations. It is proposed to use neural networks to solve large systems of ordinary differential equations. A mathematical analysis of the methods of separating a continuous medium and placing masses at points in an uneven grid of dynamic particles and a method for selecting a damping coefficient proportional to the velocity of movement of a material point is required.
4
Balabin, Helena, Antonietta Gabriella Liuzzi, Jingyuan Sun, Patrick Dupont, Rik Vanderberghe y Marie-Francine Moens. "Investigating Neural Fit Approaches for Sentence Embedding Model Paradigms". En Frontiers in Artificial Intelligence and Applications. IOS Press, 2023. http://dx.doi.org/10.3233/faia230267.
Texto completoResumen
In recent years, representations from brain activity patterns and pre-trained language models have been linked to each other based on neural fits to validate hypotheses about language processing. Nonetheless, open questions remain about what intrinsic properties of language processing these neural fits reflect and whether they differ across neural fit approaches, brain networks, and models. In this study, we use parallel sentence and functional magnetic resonance imaging data to perform a comprehensive analysis of four paradigms (masked language modeling, pragmatic coherence, semantic comparison, and contrastive learning) representing linguistic hypotheses about sentence processing. We include three sentence embedding models for each paradigm, resulting in a total of 12 models, and examine differences in their neural fit to four different brain networks using regression-based neural encoding and Representational Similarity Analysis (RSA). Among the different models tested, GPT-2, SkipThoughts, and S-RoBERTa yielded the strongest correlations with language network patterns, whereas contrastive learning-based models resulted in overall low neural fits. Our findings demonstrate that neural fits vary across brain networks and models representing the same linguistic hypothesis (e.g., GPT-2 and GPT-3). More importantly, we show the need for both neural encoding and RSA as complementary methods to provide full understanding of neural fits. All code used in the analysis is publicly available: https://github.com/lcn-kul/sentencefmricomparison.
5
Sarapisto, Teemu, Haoyu Wei, Keijo Heljanko, Arto Klami y Laura Ruotsalainen. "Subsystem Discovery in High-Dimensional Time-Series Using Masked Autoencoders". En Frontiers in Artificial Intelligence and Applications. IOS Press, 2024. http://dx.doi.org/10.3233/faia240844.
Texto completoResumen
Deep neural networks are increasingly used for time series tasks, yet they often struggle to interpretably model high-dimensional data. In this context, we consider the task of learning easy to understand connections between time-series variables, and organizing them into subsystems, directly from observed data. Our approach reconstructs multivariate time-series with a masked autoencoder, where all information between individual variables is mediated by a learned adjacency matrix. This intuitive pairwise relationship enables grouping of variables without prior knowledge of cluster quantity or size, and is particularly useful for analyzing complex sensor systems with unknown structural interdependencies. Our method simultaneously learns a useful signal representation and aids in understanding the underlying processes. We show that we can learn the correct subsystems from simulated data, and demonstrate identification of plausible subsystem structure from high-dimensional real-world data. In addition, we show that the model retains high predictive performance.
6
Udendhran, R. y Balamurugan M. "Demystification of Deep Learning-Driven Medical Image Processing and Its Impact on Future Biomedical Applications". En Deep Neural Networks for Multimodal Imaging and Biomedical Applications, 155–71. IGI Global, 2020. http://dx.doi.org/10.4018/978-1-7998-3591-2.ch010.
Texto completoResumen
The recent growth of big data has ushered in a new era of deep learning algorithms in every sphere of technological advance, including medicine, as well as in medical imaging, particularly radiology. However, the recent achievements of deep learning, in particular biomedical applications, have, to some extent, masked decades-long developments in computational technology for medical image analysis. The methods of multi-modality medical imaging have been implemented in clinical as well as research studies. Due to the reason that multi-modal image analysis and deep learning algorithms have seen fast development and provide certain benefits to biomedical applications, this chapter presents the importance of deep learning-driven medical imaging applications, future advancements, and techniques to enhance biomedical applications by employing deep learning.
7
Soujanya, R., Ravi Mohan Sharma, Manish Manish Maheshwari y Divya Prakash Shrivastava. "Fundamental Concepts in Graph Attention Networks". En Concepts and Techniques of Graph Neural Networks, 74–85. IGI Global, 2023. http://dx.doi.org/10.4018/978-1-6684-6903-3.ch006.
Texto completoResumen
Graph attention networks, also known as GATs, are a specific kind of neural network design that can function on input that is arranged as a graph. These networks make use of masked self-attentional layers in order to compensate for the shortcomings that were present in prior approaches that were based on graph convolutions. The main advantage of GAT is its ability to model the dependencies between nodes in a graph, while also allowing for different weights to be assigned to different edges in the graph. GAT is able to capture both local and global information in a graph. Local information refers to the information surrounding each node, while global information refers to the information about the entire graph. This is achieved through the use of attention mechanisms, which allow the network to selectively focus on certain nodes and edges while ignoring others. It also has scalability, interpretability, flexibility characteristics. This chapter discusses the fundamental concepts in graph attention networks.
8
Ahmed, Fahim, Md Fahim, Md Ashraful Amin, Amin Ahsan Ali y AKM Mahabubur Rahman. "Improving the Performance of Transformer-Based Models Over Classical Baselines in Multiple Transliterated Languages". En Frontiers in Artificial Intelligence and Applications. IOS Press, 2024. http://dx.doi.org/10.3233/faia240972.
Texto completoResumen
Social media users express their feelings, experiences, ideas, and stories with little or no regard for the conventions of traditional grammar. Online discourse, by its very nature, is rife with transliterated text along with code-mixing and code-switching. Transliteration is heavily featured due to the ease of inputting romanized text with standard keyboards over native scripts. Due to its ubiquity, it is a critical area of study to ensure NLP models perform well in real-world scenarios. In this paper, we analyze the performance of various language models, Tiny Large Language models, TF-IDF and Bag-of-Words feature extraction-based classical ML models, as well as zero-shot classification with ChatGPT on romanized/transliterated social media text. We chose the tasks of sentiment analysis and offensive language identification and we carried out experiments for three different languages, namely Bangla, Hindi, and Arabic, for six datasets. To our surprise, we discovered across multiple datasets that the non-neural methods perform very competitively with fine-tuned transformer-based mono/multilingual language models, tiny large language models, and ChatGPT for classification tasks in transliterated text. These classical models train in seconds using only a fraction of the computing power, and thus the carbon footprint, required by language models. We demonstrate TF-IDF and BoW-based classifiers achieve performance within around 3% of fine-tuned LMs and could thus be considered as a strong baseline for transliterated text-based NLP tasks. Additionally, we investigated various mitigation strategies such as translation and augmentation via the use of ChatGPT, as well as Masked Language Modelling to dataset-specific pretraining for language models. Depending on the dataset and language, employing those mitigation techniques yields a 2-3% further improvement in accuracy and macro-F1 above baseline.
9
Poon, Hoifung, Hai Wang y Hunter Lang. "Chapter 14. Combining Probabilistic Logic and Deep Learning for Self-Supervised Learning". En Frontiers in Artificial Intelligence and Applications. IOS Press, 2021. http://dx.doi.org/10.3233/faia210361.
Texto completoResumen
Deep learning has proven effective for various application tasks, but its applicability is limited by the reliance on annotated examples. Self-supervised learning has emerged as a promising direction to alleviate the supervision bottleneck, but existing work focuses on leveraging co-occurrences in unlabeled data for task-agnostic representation learning, as exemplified by masked language model pretraining. In this chapter, we explore task-specific self-supervision, which leverages domain knowledge to automatically annotate noisy training examples for end applications, either by introducing labeling functions for annotating individual instances, or by imposing constraints over interdependent label decisions. We first present deep probabilistic logic (DPL), which offers a unifying framework for task-specific self-supervision by composing probabilistic logic with deep learning. DPL represents unknown labels as latent variables and incorporates diverse self-supervision using probabilistic logic to train a deep neural network end-to-end using variational EM. Next, we present self-supervised self-supervision (S4), which adds to DPL the capability to learn new self-supervision automatically. Starting from an initial seed self-supervision, S4 iteratively uses the deep neural network to propose new self supervision. These are either added directly (a form of structured self-training) or verified by a human expert (as in feature-based active learning). Experiments on real-world applications such as biomedical machine reading and various text classification tasks show that task-specific self-supervision can effectively leverage domain expertise and often match the accuracy of supervised methods with a tiny fraction of human effort.
Actas de conferencias sobre el tema "Modèle en masses neurales"
1
Hernandez-Lopez, Juanita y Wilfrido Gomez-Flores. "Predicting the BI-RADS Lexicon for Mammographie Masses Using Hybrid Neural Models". En 2020 17th International Conference on Electrical Engineering, Computing Science and Automatic Control (CCE). IEEE, 2020. http://dx.doi.org/10.1109/cce50788.2020.9299155.
Texto completo
2
Atlinar, Ferhat, Tugberk Ayar, Abdurrahim Darrige, Shaza AlQays, Ahmet Bagci y Mehmet Fatih Amasyali. "Masked Word Prediction with Statistical and Neural Language Models". En 2020 Innovations in Intelligent Systems and Applications Conference (ASYU). IEEE, 2020. http://dx.doi.org/10.1109/asyu50717.2020.9259862.
Texto completo
3
Sampaio, Wener B., Edgar M. Diniz, Aristofanes C. Silva y Anselmo C. de Paiva. "Detection of Masses in Mammograms Using Cellular Neural Networks, Hidden Markov Models and Ripley's K Function". En 2009 16th International Conference on Systems, Signals and Image Processing. IEEE, 2009. http://dx.doi.org/10.1109/iwssip.2009.5367756.
Texto completo
4
Noé, Italo T., Lucas H. L. Costa y Talles H. Medeiros. "Masked Faces: Overcoming Recognition Challenges with Transfer Learning in CNNs". En Symposium on Knowledge Discovery, Mining and Learning. Sociedade Brasileira de Computação - SBC, 2023. http://dx.doi.org/10.5753/kdmile.2023.232907.
Texto completoResumen
Amidst the coronavirus pandemic, the use of masks has become one of the main ways to prevent and control the spread of this virus. However, masks impacted the performance of several face recognition models, by reducing visible features in images. The objective of this work is to present a model of convolutional neural networks with transfer learning capable of classifying thirty individuals regardless of the use of masks. The model was trained in a dataset with real images of masks and another with the insertion of computationally simulated masks and the accuracy results obtained were greater than 90%. Due to the number of classes and limitations of the dataset used, the result is consistent with the low number of related works to facial recognition with masks and highlights the complexity of the problem. It is believed that the use of a dataset with superior quality and quantity of images would make the use of the model more viable for the real world, but the tests presented are promising.
5
Le, Franck, Mudhakar Srivatsa, Krishna Kesari Reddy y Kaushik Roy. "Using Graphical Models as Explanations in Deep Neural Networks". En 2019 IEEE 16th International Conference on Mobile Ad Hoc and Sensor Systems (MASS). IEEE, 2019. http://dx.doi.org/10.1109/mass.2019.00041.
Texto completo
6
Reinders, Christoph, Frederik Schubert y Bodo Rosenhahn. "ChimeraMix: Image Classification on Small Datasets via Masked Feature Mixing". En Thirty-First International Joint Conference on Artificial Intelligence {IJCAI-22}. California: International Joint Conferences on Artificial Intelligence Organization, 2022. http://dx.doi.org/10.24963/ijcai.2022/181.
Texto completoResumen
Deep convolutional neural networks require large amounts of labeled data samples. For many real-world applications, this is a major limitation which is commonly treated by augmentation methods. In this work, we address the problem of learning deep neural networks on small datasets. Our proposed architecture called ChimeraMix learns a data augmentation by generating compositions of instances. The generative model encodes images in pairs, combines the features guided by a mask, and creates new samples. For evaluation, all methods are trained from scratch without any additional data. Several experiments on benchmark datasets, e.g. ciFAIR-10, STL-10, and ciFAIR-100, demonstrate the superior performance of ChimeraMix compared to current state-of-the-art methods for classification on small datasets. Code is available at https://github.com/creinders/ChimeraMix.
7
Guo, Quan, Hossein Rajaby Faghihi, Yue Zhang, Andrzej Uszok y Parisa Kordjamshidi. "Inference-Masked Loss for Deep Structured Output Learning". En Twenty-Ninth International Joint Conference on Artificial Intelligence and Seventeenth Pacific Rim International Conference on Artificial Intelligence {IJCAI-PRICAI-20}. California: International Joint Conferences on Artificial Intelligence Organization, 2020. http://dx.doi.org/10.24963/ijcai.2020/382.
Texto completoResumen
Structured learning algorithms usually involve an inference phase that selects the best global output variables assignments based on the local scores of all possible assignments. We extend deep neural networks with structured learning to combine the power of learning representations and leveraging the use of domain knowledge in the form of output constraints during training. Introducing a non-differentiable inference module to gradient-based training is a critical challenge. Compared to using conventional loss functions that penalize every local error independently, we propose an inference-masked loss that takes into account the effect of inference and does not penalize the local errors that can be corrected by the inference. We empirically show the inference-masked loss combined with the negative log-likelihood loss improves the performance on different tasks, namely entity relation recognition on CoNLL04 and ACE2005 corpora, and spatial role labeling on CLEF 2017 mSpRL dataset. We show the proposed approach helps to achieve better generalizability, particularly in the low-data regime.
8
Shi, Yunsheng, Zhengjie Huang, Shikun Feng, Hui Zhong, Wenjing Wang y Yu Sun. "Masked Label Prediction: Unified Message Passing Model for Semi-Supervised Classification". En Thirtieth International Joint Conference on Artificial Intelligence {IJCAI-21}. California: International Joint Conferences on Artificial Intelligence Organization, 2021. http://dx.doi.org/10.24963/ijcai.2021/214.
Texto completoResumen
Graph neural network (GNN) and label propagation algorithm (LPA) are both message passing algorithms, which have achieved superior performance in semi-supervised classification. GNN performs feature propagation by a neural network to make predictions, while LPA uses label propagation across graph adjacency matrix to get results. However, there is still no effective way to directly combine these two kinds of algorithms. To address this issue, we propose a novel Unified Message Passaging Model (UniMP) that can incorporate feature and label propagation at both training and inference time. First, UniMP adopts a Graph Transformer network, taking feature embedding and label embedding as input information for propagation. Second, to train the network without overfitting in self-loop input label information, UniMP introduces a masked label prediction strategy, in which some percentage of input label information are masked at random, and then predicted. UniMP conceptually unifies feature propagation and label propagation and is empirically powerful. It obtains new state-of-the-art semi-supervised classification results in Open Graph Benchmark (OGB).
9
Nishimoto, Hiroyuki. "Effective deep learning through bidirectional reading on masked language model". En Human Systems Engineering and Design (IHSED 2021) Future Trends and Applications. AHFE International, 2021. http://dx.doi.org/10.54941/ahfe1001178.
Texto completoResumen
Google BERT is a neural network that is good at natural language processing. It has two major strategies. One is “Masked language Model” to clear the word-level relationships, and the other is “Next Sentence Prediction” to clear sentence-level relationships. In the masked language model, with the task of masking some words in sentences, BERT learns to predict the original word from context. Some questions come to mind. Why BERT achieves effective learning by reading in two ways from fore and back? What is the difference between bidirectional reading? BERT learns to predict the original word using the surrounding words as context and to make two-way predictions by forward and backward readings in order to increase the precision. Besides, the bidirectional reading technique can be applied to scenario planning especially using back-casting from the future. This paper clarifies these mechanisms.
10
Kelly, Sean T., Andrea Lupini y Bogdan I. Epureanu. "Data-Driven Approach for Identifying Mistuning in As-Manufactured Blisks". En ASME Turbo Expo 2021: Turbomachinery Technical Conference and Exposition. American Society of Mechanical Engineers, 2021. http://dx.doi.org/10.1115/gt2021-59887.
Texto completoResumen
Abstract Sector-to-sector geometry or material property variations in as-manufactured bladed disks, or blisks, can result in significantly greater vibration responses during operation compared to nominally cyclic symmetric designs. The dynamics of blisks are sensitive to these unavoidable deviations, known as mistuning, making the identification of mistuning in as-manufactured blisks necessary for accurately predicting their vibration. As in previous mistuning modeling and identification approaches, the mistuning of interest is small and is parameterized by using deviations in cantilever blade-alone frequencies. Such mistuning parameterization is popular because it can be applied through blade-to-blade stiffness deviations in computational reduced-order models used to predict blisk dynamics. Previous approaches to identify such mistuning parameters often require the identification of modal information or blade-isolation techniques such blade detuning using masses or adding damping pads. However, modal information can be difficult to obtain accurately even in optimal bench conditions. Additionally, in practice it can be difficult to isolate individual blades by restricting blade motion around the blisk or detuning individual blades through added masses due to geometric constraints. In this paper, we present a method for mistuning identification using a data-driven approach based on a neural network. The network is first trained using surrogate computational data. Thus, the data-driven portion of the approach is executed using surrogate computational methods. With the trained network, mistuning in all sectors of blisks with the same nominal geometry can be identified by using a small number of forced responses and the forcing phase information from traveling-wave excitation. In this approach, no system or sector-level modal response information, restrictive blade isolation, or mass detuning are required. We additionally present a method for forcing frequency selection and response conditioning to improve identification accuracy. Validation of this approach is presented using a finite element blisk model containing stiffness mistuning within the blades to create computationally generated surrogate data. It is shown that mistuning can be predicted accurately using forced responses containing a significant amount of absolute and relative measurement noise, mimicking responses collected from experimental measurements. In addition, it is shown that mistuning can be predicted independently and accurately using different engine orders of excitation in regions of high modal density.
Informes sobre el tema "Modèle en masses neurales"
1
Semerikov, Serhiy O., Illia O. Teplytskyi, Yuliia V. Yechkalo y Arnold E. Kiv. Computer Simulation of Neural Networks Using Spreadsheets: The Dawn of the Age of Camelot. [б. в.], noviembre de 2018. http://dx.doi.org/10.31812/123456789/2648.
Texto completoResumen
The article substantiates the necessity to develop training methods of computer simulation of neural networks in the spreadsheet environment. The systematic review of their application to simulating artificial neural networks is performed. The authors distinguish basic approaches to solving the problem of network computer simulation training in the spreadsheet environment, joint application of spreadsheets and tools of neural network simulation, application of third-party add-ins to spreadsheets, development of macros using the embedded languages of spreadsheets; use of standard spreadsheet add-ins for non-linear optimization, creation of neural networks in the spreadsheet environment without add-ins and macros. After analyzing a collection of writings of 1890-1950, the research determines the role of the scientific journal “Bulletin of Mathematical Biophysics”, its founder Nicolas Rashevsky and the scientific community around the journal in creating and developing models and methods of computational neuroscience. There are identified psychophysical basics of creating neural networks, mathematical foundations of neural computing and methods of neuroengineering (image recognition, in particular). The role of Walter Pitts in combining the descriptive and quantitative theories of training is discussed. It is shown that to acquire neural simulation competences in the spreadsheet environment, one should master the models based on the historical and genetic approach. It is indicated that there are three groups of models, which are promising in terms of developing corresponding methods – the continuous two-factor model of Rashevsky, the discrete model of McCulloch and Pitts, and the discrete-continuous models of Householder and Landahl.
2
Fessel, Kimberly. Machine Learning in Python. Instats Inc., 2024. http://dx.doi.org/10.61700/s74zy0ivgwioe1764.
Texto completoResumen
This intensive, hands-on workshop offers a deep dive into machine learning with Python, designed for PhD students, professors, and researchers across various fields. Participants will master practical skills in data cleaning, exploratory data analysis, and building powerful machine learning models, including neural networks, to elevate their research. With real-world coding exercises and expert guidance, this workshop will equip you with the tools to turn data into actionable insights.
3
Engel, Bernard, Yael Edan, James Simon, Hanoch Pasternak y Shimon Edelman. Neural Networks for Quality Sorting of Agricultural Produce. United States Department of Agriculture, julio de 1996. http://dx.doi.org/10.32747/1996.7613033.bard.
Texto completoResumen
The objectives of this project were to develop procedures and models, based on neural networks, for quality sorting of agricultural produce. Two research teams, one in Purdue University and the other in Israel, coordinated their research efforts on different aspects of each objective utilizing both melons and tomatoes as case studies. At Purdue: An expert system was developed to measure variances in human grading. Data were acquired from eight sensors: vision, two firmness sensors (destructive and nondestructive), chlorophyll from fluorescence, color sensor, electronic sniffer for odor detection, refractometer and a scale (mass). Data were analyzed and provided input for five classification models. Chlorophyll from fluorescence was found to give the best estimation for ripeness stage while the combination of machine vision and firmness from impact performed best for quality sorting. A new algorithm was developed to estimate and minimize training size for supervised classification. A new criteria was established to choose a training set such that a recurrent auto-associative memory neural network is stabilized. Moreover, this method provides for rapid and accurate updating of the classifier over growing seasons, production environments and cultivars. Different classification approaches (parametric and non-parametric) for grading were examined. Statistical methods were found to be as accurate as neural networks in grading. Classification models by voting did not enhance the classification significantly. A hybrid model that incorporated heuristic rules and either a numerical classifier or neural network was found to be superior in classification accuracy with half the required processing of solely the numerical classifier or neural network. In Israel: A multi-sensing approach utilizing non-destructive sensors was developed. Shape, color, stem identification, surface defects and bruises were measured using a color image processing system. Flavor parameters (sugar, acidity, volatiles) and ripeness were measured using a near-infrared system and an electronic sniffer. Mechanical properties were measured using three sensors: drop impact, resonance frequency and cyclic deformation. Classification algorithms for quality sorting of fruit based on multi-sensory data were developed and implemented. The algorithms included a dynamic artificial neural network, a back propagation neural network and multiple linear regression. Results indicated that classification based on multiple sensors may be applied in real-time sorting and can improve overall classification. Advanced image processing algorithms were developed for shape determination, bruise and stem identification and general color and color homogeneity. An unsupervised method was developed to extract necessary vision features. The primary advantage of the algorithms developed is their ability to learn to determine the visual quality of almost any fruit or vegetable with no need for specific modification and no a-priori knowledge. Moreover, since there is no assumption as to the type of blemish to be characterized, the algorithm is capable of distinguishing between stems and bruises. This enables sorting of fruit without knowing the fruits' orientation. A new algorithm for on-line clustering of data was developed. The algorithm's adaptability is designed to overcome some of the difficulties encountered when incrementally clustering sparse data and preserves information even with memory constraints. Large quantities of data (many images) of high dimensionality (due to multiple sensors) and new information arriving incrementally (a function of the temporal dynamics of any natural process) can now be processed. Furhermore, since the learning is done on-line, it can be implemented in real-time. The methodology developed was tested to determine external quality of tomatoes based on visual information. An improved model for color sorting which is stable and does not require recalibration for each season was developed for color determination. Excellent classification results were obtained for both color and firmness classification. Results indicted that maturity classification can be obtained using a drop-impact and a vision sensor in order to predict the storability and marketing of harvested fruits. In conclusion: We have been able to define quantitatively the critical parameters in the quality sorting and grading of both fresh market cantaloupes and tomatoes. We have been able to accomplish this using nondestructive measurements and in a manner consistent with expert human grading and in accordance with market acceptance. This research constructed and used large databases of both commodities, for comparative evaluation and optimization of expert system, statistical and/or neural network models. The models developed in this research were successfully tested, and should be applicable to a wide range of other fruits and vegetables. These findings are valuable for the development of on-line grading and sorting of agricultural produce through the incorporation of multiple measurement inputs that rapidly define quality in an automated manner, and in a manner consistent with the human graders and inspectors.
4
Roberson, Madeleine, Kathleen Inman, Ashley Carey, Isaac Howard y Jameson Shannon. Probabilistic neural networks that predict compressive strength of high strength concrete in mass placements using thermal history. Engineer Research and Development Center (U.S.), junio de 2022. http://dx.doi.org/10.21079/11681/44483.
Texto completoResumen
This study explored the use of artificial neural networks to predict UHPC compressive strengths given thermal history and key mix components. The model developed herein employs Bayesian variational inference using Monte Carlo dropout to convey prediction uncertainty using 735 datapoints on seven UHPC mixtures collected using a variety of techniques. Datapoints contained a measured compressive strength along with three curing inputs (specimen maturity, maximum temperature experienced during curing, time of maximum temperature) and five mixture inputs to distinguish each UHPC mixture (cement type, silicon dioxide content, mix type, water to cementitious material ratio, and admixture dosage rate). Input analysis concluded that predictions were more sensitive to curing inputs than mixture inputs. On average, 8.2% of experimental results in the final model fell outside of the predicted range with 67.9%of these cases conservatively underpredicting. The results support that this model methodology is able to make sufficient probabilistic predictions within the scope of the provided dataset but is not for extrapolating beyond the training data. In addition, the model was vetted using various datasets obtained from literature to assess its versatility. Overall this model is a promising advancement towards predicting mechanical properties of high strength concrete with known uncertainties.
5
Altstein, Miriam y Ronald Nachman. Rationally designed insect neuropeptide agonists and antagonists: application for the characterization of the pyrokinin/Pban mechanisms of action in insects. United States Department of Agriculture, octubre de 2006. http://dx.doi.org/10.32747/2006.7587235.bard.
Texto completoResumen
The general objective of this BARD project focused on rationally designed insect neuropeptide (NP) agonists and antagonists, their application for the characterization of the mechanisms of action of the pyrokinin/PBAN (PK-PBAN) family and the development of biostable, bioavailable versions that can provide the basis for development of novel, environmentally-friendly pest insect control agents. The specific objectives of the study, as originally proposed, were to: (i) Test stimulatory potencies of rationally designed backbone cyclic (BBC) peptides on pheromonotropic, melanotropic, myotropic and pupariation activities; (ii) Test the inhibitory potencies of the BBC compounds on the above activities evoked either by synthetic peptides (PBAN, LPK, myotropin and pheromonotropin) or by the natural endogenous mechanism; (iii) Determine the bioavailability of the most potent BBC compounds that will be found in (ii); (iv) Design, synthesize and examine novel PK/PBAN analogs with enhanced bioavailability and receptor binding; (v) Design and synthesize ‘magic bullet’ analogs and examine their ability to selectively kill cells expressing the PK/PBAN receptor. To achieve these goals the agonistic and antagonistic activities/properties of rationally designed linear and BBC neuropeptide (NP) were thoroughly studied and the information obtained was further used for the design and synthesis of improved compounds toward the design of an insecticide prototype. The study revealed important information on the structure activity relationship (SAR) of agonistic/antagonistic peptides, including definitive identification of the orientation of the Pro residue as trans for agonist activity in 4 PK/PBANbioassays (pheromonotropic, pupariation, melanotropic, & hindgut contractile) and a PK-related CAP₂b bioassay (diuretic); indications that led to the identification of a novel scaffold to develop biostbiostable, bioavailable peptidomimetic PK/PBANagonists/antagonists. The work led to the development of an arsenal of PK/PBAN antagonists with a variety of selectivity profiles; whether between different PKbioassays, or within the same bioassay between different natural elicitors. Examples include selective and non-selective BBC and novel amphiphilic PK pheromonotropic and melanotropic antagonists some of which are capable of penetrating the moth cuticle in efficacious quantities. One of the latter analog group demonstrated unprecedented versatility in its ability to antagonize a broad spectrum of pheromonotropic elicitors. A novel, transPro mimetic motif was proposed & used to develop a strong, selective PK agonist of the melanotropic bioassay in moths. The first antagonist (pure) of PK-related CAP₂b diuresis in flies was developed using a cisPro mimetic motif; an indication that while a transPro orientation is associated with receptor agonism, a cisPro orientation is linked with an antagonist interaction. A novel, biostablePK analog, incorporating β-amino acids at key peptidase-susceptible sites, exhibited in vivo pheromonotropic activity that by far exceeded that of PBAN when applied topically. Direct analysis of neural tissue by state-of-the-art MALDI-TOF/TOF mass spectrometry was used to identify specific PK/PK-related peptides native to eight arthropod pest species [house (M. domestica), stable (S. calcitrans), horn (H. irritans) & flesh (N. bullata) flies; Southern cattle fever tick (B. microplus), European tick (I. ricinus), yellow fever mosquito (A. aegypti), & Southern Green Stink Bug (N. viridula)]; including the unprecedented identification of mass-identical Leu/Ile residues and the first identification of NPs from a tick or the CNS of Hemiptera. Evidence was obtained for the selection of Neb-PK-2 as the primary pupariation factor of the flesh fly (N. bullata) among native PK/PK-related candidates. The peptidomic techniques were also used to map the location of PK/PK-related NP in the nervous system of the model fly D. melanogaster. Knowledge of specific PK sequences can aid in the future design of species specific (or non-specific) NP agonists/antagonists. In addition, the study led to the first cloning of a PK/PBAN receptor from insect larvae (S. littoralis), providing the basis for SAR analysis for the future design of 2ⁿᵈgeneration selective and/or nonselective agonists/antagonists. Development of a microplate ligand binding assay using the PK/PBAN pheromone gland receptor was also carried out. The assay will enable screening, including high throughput, of various libraries (chemical, molecular & natural product) for the discovery of receptor specific agonists/antagonists. In summary, the body of work achieves several key milestones and brings us significantly closer to the development of novel, environmentally friendly pest insect management agents based on insect PK/PBANNPs capable of disrupting critical NP-regulated functions.
6
Galili, Naftali, Roger P. Rohrbach, Itzhak Shmulevich, Yoram Fuchs y Giora Zauberman. Non-Destructive Quality Sensing of High-Value Agricultural Commodities Through Response Analysis. United States Department of Agriculture, octubre de 1994. http://dx.doi.org/10.32747/1994.7570549.bard.
Texto completoResumen
The objectives of this project were to develop nondestructive methods for detection of internal properties and firmness of fruits and vegetables. One method was based on a soft piezoelectric film transducer developed in the Technion, for analysis of fruit response to low-energy excitation. The second method was a dot-matrix piezoelectric transducer of North Carolina State University, developed for contact-pressure analysis of fruit during impact. Two research teams, one in Israel and the other in North Carolina, coordinated their research effort according to the specific objectives of the project, to develop and apply the two complementary methods for quality control of agricultural commodities. In Israel: An improved firmness testing system was developed and tested with tropical fruits. The new system included an instrumented fruit-bed of three flexible piezoelectric sensors and miniature electromagnetic hammers, which served as fruit support and low-energy excitation device, respectively. Resonant frequencies were detected for determination of firmness index. Two new acoustic parameters were developed for evaluation of fruit firmness and maturity: a dumping-ratio and a centeroid of the frequency response. Experiments were performed with avocado and mango fruits. The internal damping ratio, which may indicate fruit ripeness, increased monotonically with time, while resonant frequencies and firmness indices decreased with time. Fruit samples were tested daily by destructive penetration test. A fairy high correlation was found in tropical fruits between the penetration force and the new acoustic parameters; a lower correlation was found between this parameter and the conventional firmness index. Improved table-top firmness testing units, Firmalon, with data-logging system and on-line data analysis capacity have been built. The new device was used for the full-scale experiments in the next two years, ahead of the original program and BARD timetable. Close cooperation was initiated with local industry for development of both off-line and on-line sorting and quality control of more agricultural commodities. Firmalon units were produced and operated in major packaging houses in Israel, Belgium and Washington State, on mango and avocado, apples, pears, tomatoes, melons and some other fruits, to gain field experience with the new method. The accumulated experimental data from all these activities is still analyzed, to improve firmness sorting criteria and shelf-life predicting curves for the different fruits. The test program in commercial CA storage facilities in Washington State included seven apple varieties: Fuji, Braeburn, Gala, Granny Smith, Jonagold, Red Delicious, Golden Delicious, and D'Anjou pear variety. FI master-curves could be developed for the Braeburn, Gala, Granny Smith and Jonagold apples. These fruits showed a steady ripening process during the test period. Yet, more work should be conducted to reduce scattering of the data and to determine the confidence limits of the method. Nearly constant FI in Red Delicious and the fluctuations of FI in the Fuji apples should be re-examined. Three sets of experiment were performed with Flandria tomatoes. Despite the complex structure of the tomatoes, the acoustic method could be used for firmness evaluation and to follow the ripening evolution with time. Close agreement was achieved between the auction expert evaluation and that of the nondestructive acoustic test, where firmness index of 4.0 and more indicated grade-A tomatoes. More work is performed to refine the sorting algorithm and to develop a general ripening scale for automatic grading of tomatoes for the fresh fruit market. Galia melons were tested in Israel, in simulated export conditions. It was concluded that the Firmalon is capable of detecting the ripening of melons nondestructively, and sorted out the defective fruits from the export shipment. The cooperation with local industry resulted in development of automatic on-line prototype of the acoustic sensor, that may be incorporated with the export quality control system for melons. More interesting is the development of the remote firmness sensing method for sealed CA cool-rooms, where most of the full-year fruit yield in stored for off-season consumption. Hundreds of ripening monitor systems have been installed in major fruit storage facilities, and being evaluated now by the consumers. If successful, the new method may cause a major change in long-term fruit storage technology. More uses of the acoustic test method have been considered, for monitoring fruit maturity and harvest time, testing fruit samples or each individual fruit when entering the storage facilities, packaging house and auction, and in the supermarket. This approach may result in a full line of equipment for nondestructive quality control of fruits and vegetables, from the orchard or the greenhouse, through the entire sorting, grading and storage process, up to the consumer table. The developed technology offers a tool to determine the maturity of the fruits nondestructively by monitoring their acoustic response to mechanical impulse on the tree. A special device was built and preliminary tested in mango fruit. More development is needed to develop a portable, hand operated sensing method for this purpose. In North Carolina: Analysis method based on an Auto-Regressive (AR) model was developed for detecting the first resonance of fruit from their response to mechanical impulse. The algorithm included a routine that detects the first resonant frequency from as many sensors as possible. Experiments on Red Delicious apples were performed and their firmness was determined. The AR method allowed the detection of the first resonance. The method could be fast enough to be utilized in a real time sorting machine. Yet, further study is needed to look for improvement of the search algorithm of the methods. An impact contact-pressure measurement system and Neural Network (NN) identification method were developed to investigate the relationships between surface pressure distributions on selected fruits and their respective internal textural qualities. A piezoelectric dot-matrix pressure transducer was developed for the purpose of acquiring time-sampled pressure profiles during impact. The acquired data was transferred into a personal computer and accurate visualization of animated data were presented. Preliminary test with 10 apples has been performed. Measurement were made by the contact-pressure transducer in two different positions. Complementary measurements were made on the same apples by using the Firmalon and Magness Taylor (MT) testers. Three-layer neural network was designed. 2/3 of the contact-pressure data were used as training input data and corresponding MT data as training target data. The remaining data were used as NN checking data. Six samples randomly chosen from the ten measured samples and their corresponding Firmalon values were used as the NN training and target data, respectively. The remaining four samples' data were input to the NN. The NN results consistent with the Firmness Tester values. So, if more training data would be obtained, the output should be more accurate. In addition, the Firmness Tester values do not consistent with MT firmness tester values. The NN method developed in this study appears to be a useful tool to emulate the MT Firmness test results without destroying the apple samples. To get more accurate estimation of MT firmness a much larger training data set is required. When the larger sensitive area of the pressure sensor being developed in this project becomes available, the entire contact 'shape' will provide additional information and the neural network results would be more accurate. It has been shown that the impact information can be utilized in the determination of internal quality factors of fruit. Until now,