Tesi: "Adaptation de domaines"

1

Fernandes, Montesuma Eduardo. "Multi-Source Domain Adaptation through Wasserstein Barycenters". Electronic Thesis or Diss., université Paris-Saclay, 2024. http://www.theses.fr/2024UPASG045.

Testo completo

Abstract (sommario):

Les systèmes d'apprentissage automatique fonctionnent sous l'hypothèse que les conditions d'entraînement et de test ne changent pas. Néanmoins, cette hypothèse est rarement vérifiée en pratique. En conséquence, le système est entraîné avec des données qui ne sont plus représentatives des données sur lesquelles il sera testé : la mesure de probabilité des données évolue entre les périodes d'entraînement et de test. Ce scénario est connu dans la littérature sous le nom de décalage de distribution entre deux domaines : une source et une cible. Une généralisation évidente de ce problème considère que les données d'entraînement présentent elles-mêmes plusieurs décalages intrinsèques. On parle, donc, d'adaptation de domaine à sources multiples (MSDA). Dans ce contexte, le transport optimal est un outil de mathématique utile. En particulier, qui sert pour comparer et manipuler des mesures de probabilité. Cette thèse étudie les contributions du transport optimal à l'adaptation de domaines à sources multiples. Nous le faisons à travers des barycentres de Wasserstein, un objet qui définit une moyenne pondérée, dans l'espace des mesures de probabilité, des multiples domaines en MSDA. Basé sur ce concept, nous proposons : (i) une nouvelle notion de barycentre lorsque les mesures en question sont étiquetées, (ii) un nouveau problème d'apprentissage de dictionnaire sur des mesures de probabilité empiriques et (iii) de nouveaux outils pour l'adaptation de domaines via le transport optimal de modèles de mélanges Gaussiens. Nos méthodes améliorent les performances de l'adaptation de domaines par rapport aux méthodes existantes utilisant le transport optimal sur des benchmarks d'images et de diagnostic de défauts inter-domaines. Notre travail ouvre une perspective de recherche intéressante sur l'apprentissage de l'enveloppe barycentrique de mesures de probabilité
Machine learning systems work under the assumption that training and test conditions are uniform, i.e., they do not change. However, this hypothesis is seldom met in practice. Hence, the system is trained with data that is no longer representative of the data it will be tested on. This case is represented by a shift in the probability measure generating the data. This scenario is known in the literature as distributional shift between two domains: a source, and a target. A straightforward generalization of this problem is when training data itself exhibit shifts on its own. In this case, one consider Multi Source Domain Adaptation (MSDA). In this context, optimal transport is an useful field of mathematics. Especially, optimal transport serves as a toolbox, for comparing and manipulating probability measures. This thesis studies the contributions of optimal transport to multi-source domain adaptation. We do so through Wasserstein barycenters, an object that defines a weighted average, in the space of probability measures, for the multiple domains in MSDA. Based on this concept, we propose: (i) a novel notion of barycenter, when the measures at hand are equipped with labels, (ii) a novel dictionary learning problem over empirical probability measures and (iii) new tools for domain adaptation through the optimal transport of Gaussian mixture models. Through our methods, we are able to improve domain adaptation performance in comparison with previous optimal transport-based methods on image, and cross-domain fault diagnosis benchmarks. Our work opens an interesting research direction, on learning the barycentric hull of probability measures

Gli stili APA, Harvard, Vancouver, ISO e altri

2

Lévesque-Gravel, Anick. "Adaptation de la formule de Schwarz-Christoffel aux domaines multiplement connexes". Master's thesis, Université Laval, 2015. http://hdl.handle.net/20.500.11794/26169.

Testo completo

Abstract (sommario):

Tableau d'honneur de la Faculté des études supérieures et postdoctorales, 2015-2016
La formule de Schwarz–Christoffel permet de trouver une transformation conforme entre un domaine polygonal et un disque. Par contre, cette formule ne s’applique qu’aux domaines simplement connexes. Récemment, Darren Crowdy a obtenu une généralisation de cette formule pour les domaines multiplement connexes. Celle-ci envoie des domaines circulaires sur des domaines polygonaux. Ce mémoire vise à faire la démonstration de la formule développée par Crowdy. Pour ce faire, il faudra définir la fonction de Schottky–Klein ainsi que la fonction de Green modifiée. Il faudra aussi introduire les domaines canoniques.

Gli stili APA, Harvard, Vancouver, ISO e altri

3

Meftah, Sara. "Neural Transfer Learning for Domain Adaptation in Natural Language Processing". Thesis, université Paris-Saclay, 2021. http://www.theses.fr/2021UPASG021.

Testo completo

Abstract (sommario):

Les méthodes d’apprentissage automatique qui reposent sur les Réseaux de Neurones (RNs) ont démontré des performances de prédiction qui s'approchent de plus en plus de la performance humaine dans plusieurs applications du Traitement Automatique de la Langue (TAL) qui bénéficient de la capacité des différentes architectures des RNs à généraliser à partir des régularités apprises à partir d'exemples d'apprentissage. Toutefois, ces modèles sont limités par leur dépendance aux données annotées. En effet, pour être performants, ces modèles neuronaux ont besoin de corpus annotés de taille importante. Par conséquent, uniquement les langues bien dotées peuvent bénéficier directement de l'avancée apportée par les RNs, comme par exemple les formes formelles des langues. Dans le cadre de cette thèse, nous proposons des méthodes d'apprentissage par transfert neuronal pour la construction d'outils de TAL pour les langues peu dotées en exploitant leurs similarités avec des langues bien dotées. Précisément, nous expérimentons nos approches pour le transfert à partir du domaine source des textes formels vers le domaine cible des textes informels (langue utilisée dans les réseaux sociaux). Tout au long de cette thèse nous proposons différentes contributions. Tout d'abord, nous proposons deux approches pour le transfert des connaissances encodées dans les représentations neuronales d'un modèle source, pré-entraîné sur les données annotées du domaine source, vers un modèle cible, adapté par la suite sur quelques exemples annotés du domaine cible. La première méthode transfère des représentations contextuelles pré-entraînées sur le domaine source. Tandis que la deuxième méthode utilise des poids pré-entraînés pour initialiser les paramètres du modèle cible. Ensuite, nous effectuons une série d'analyses pour repérer les limites des méthodes proposées ci-dessus. Nous constatons que, même si l'approche d'apprentissage par transfert proposée améliore les résultats du domaine cible, un transfert négatif « dissimulé » peut atténuer le gain final apporté par l'apprentissage par transfert. De plus, une analyse interprétative du modèle pré-entraîné, montre que les neurones pré-entraînés peuvent être biaisés par ce qu'ils ont appris du domaine source, et donc peuvent avoir des difficultés à apprendre des « patterns » spécifiques au domaine cible. Issu de notre analyse, nous proposons un nouveau schéma d'adaptation qui augmente le modèle cible avec des neurones normalisés, pondérés et initialisés aléatoirement qui permettent une meilleure adaptation au domaine cible tout en conservant les connaissances apprises du domaine source. Enfin, nous proposons une approche d’apprentissage par transfert qui permet de profiter des similarités entre différentes tâches, en plus des connaissances pré-apprises du domaine source
Recent approaches based on end-to-end deep neural networks have revolutionised Natural Language Processing (NLP), achieving remarkable results in several tasks and languages. Nevertheless, these approaches are limited with their "gluttony" in terms of annotated data, since they rely on a supervised training paradigm, i.e. training from scratch on large amounts of annotated data. Therefore, there is a wide gap between NLP technologies capabilities for high-resource languages compared to the long tail of low-resourced languages. Moreover, NLP researchers have focused much of their effort on training NLP models on the news domain, due to the availability of training data. However, many research works have highlighted that models trained on news fail to work efficiently on out-of-domain data, due to their lack of robustness against domain shifts. This thesis presents a study of transfer learning approaches, through which we propose different methods to take benefit from the pre-learned knowledge on the high-resourced domain to enhance the performance of neural NLP models in low-resourced settings. Precisely, we apply our approaches to transfer from the news domain to the social media domain. Indeed, despite the importance of its valuable content for a variety of applications (e.g. public security, health monitoring, or trends highlight), this domain is still poor in terms of annotated data. We present different contributions. First, we propose two methods to transfer the knowledge encoded in the neural representations of a source model pretrained on large labelled datasets from the source domain to the target model, further adapted by a fine-tuning on few annotated examples from the target domain. The first transfers contextualised supervisedly pretrained representations, while the second method transfers pretrained weights, used to initialise the target model's parameters. Second, we perform a series of analysis to spot the limits of the above-mentioned proposed methods. We find that even if the proposed transfer learning approach enhances the performance on social media domain, a hidden negative transfer may mitigate the final gain brought by transfer learning. In addition, an interpretive analysis of the pretrained model, show that pretrained neurons may be biased by what they have learned from the source domain, thus struggle with learning uncommon target-specific patterns. Third, stemming from our analysis, we propose a new adaptation scheme which augments the target model with normalised, weighted and randomly initialised neurons that beget a better adaptation while maintaining the valuable source knowledge. Finally, we propose a model, that in addition to the pre-learned knowledge from the high-resource source-domain, takes advantage of various supervised NLP tasks

Gli stili APA, Harvard, Vancouver, ISO e altri

4

Marchand, Morgane. "Domaines et fouille d'opinion : une étude des marqueurs multi-polaires au niveau du texte". Thesis, Paris 11, 2015. http://www.theses.fr/2015PA112026/document.

Testo completo

Abstract (sommario):

Cette thèse s’intéresse à l’adaptation d’un classifieur statistique d’opinion au niveau du texte d’un domaine à un autre. Cependant, nous exprimons notre opinion différemment selon ce dont nous parlons. Un même mot peut ne pas désigner pas la même chose ou bien ne pas avoir la même connotation selon le thème de la discussion. Si ces mots ne sont pas détectés, ils induiront des erreurs de classification.Nous appelons donc marqueurs multi-polaires des mots ou bigrammes dont la présence indique une certaine polarité du texte entier, différente selon le domaine du texte. Cette thèse est consacrées à leur étude. Ces marqueurs sont détectés à l’aide d’un test du khi2 lorsque l’on dispose d’annotations au niveau du texte dans les deux domaines d’intérêt. Nous avons également proposé une méthode de détection semi-supervisé. Nous utilisons une collections de mots pivots auto-épurés afin d’assurer une polarité stable d’un domaine à un autre.Nous avons également vérifié la pertinence linguistique des mots sélectionnés en organisant une campagne d’annotation manuelle. Les mots ainsi validés comme multi-polaires peuvent être des éléments de contexte, des mots exprimant ou expliquant une opinion ou bien désignant l’objet sur lequel l’opinion est portée. Notre étude en contexte a également mis en lumière trois causes principale de changement de polarité : le changement de sens, le changement d’objet et le changement d’utilisation.Pour finir, nous avons étudié l’influence de la détection des marqueurs multi-polaires sur la classification de l’opinion au niveau du texte par des classifieurs automatiques dans trois cas distincts : adaptation d’un domaine source à un domaine cible, corpus multi-domaine, corpus en domaine ouvert. Les résultats de ces expériences montrent que plus le transfert initial est difficile, plus la prise en compte des marqueurs multi-polaires peut améliorer la classification, allant jusqu’à plus cinq points d’exactitude
In this thesis, we are studying the adaptation of a text level opinion classifier across domains. Howerver, people express their opinion in a different way depending on the subject of the conversation. The same word in two different domains can refer to different objects or have an other connotation. If these words are not detected, they will lead to classification errors.We call these words or bigrams « multi-polarity marquers ». Their presence in a text signals a polarity wich is different according to the domain of the text. Their study is the subject of this thesis. These marquers are detected using a khi2 test if labels exist in both targeted domains. We also propose a semi-supervised detection method for the case with labels in only one domain. We use a collection of auto-epurated pivot words in order to assure a stable polarity accross domains.We have also checked the linguistic interest of the selected words with a manual evaluation campaign. The validated words can be : a word of context, a word giving an opinion, a word explaining an opinion or a word wich refer to the evaluated object. Our study also show that the causes of the changing polarity are of three kinds : changing meaning, changing object or changing use.Finally, we have studyed the influence of multi-polarity marquers on opinion classification at text level in three different cases : adaptation of a source domain to a target domain, multi-domain corpora and open domain corpora. The results of our experiments show that the potential improvement is bigger when the initial transfer was difficult. In the favorable cases, we improve accurracy up to five points

Gli stili APA, Harvard, Vancouver, ISO e altri

5

Passerieux, Emilie. "Corrélation entre l'organisation spatiale du perimysium et des domaines subcellulaires des fibres musculaires squelettiques : implication dans la transmission latérale des forces et conséquences possibles sur les adaptations du muscle à l'exercice physique". Bordeaux 2, 2006. http://www.theses.fr/2006BOR21358.

Testo completo

Abstract (sommario):

Nous avons exploré la possibilité qu’une fraction du tissu conjonctif intramusculaire, le périmysium soit impliqué dans les mécanismes qui sont à l’origine des adaptations musculaires. Pour cela, nous avons montré dans le muscle squelettique Flexor carpi radialis de bœuf que le périmysium transmet latéralement les forces de la contraction musculaire depuis les fibres musculaires jusqu’aux tendons. D’autre part, nous avons montré que sa répartition spatiale correspond à la surface des fibres musculaires à la répartition des intégrines et à la présence des cellules satellites et dans les fibres musculaires à la répartition des noyaux, des mitochondries subsarcolemmales et des myosines. Ainsi l’association périmysium – fibre musculaire constitue l’origine d’un système mécanosensible apte à expliquer les adaptations à court et long terme et nos premières investigations peuvent servir de point de départ pour explorer les voies par lesquelles s’expliquent les adaptations musculaires
We investigated the possibility that the perimysium, a component of intramuscular connective tissue, is involved in muscular adaptation mechanisms. We demonstrated in bovine skeletal Flexor carpi radialis muscle that (i) the perimysium drives the forces of muscular contraction from myofibers to the tendons, (ii) the spatial distribution of the perimysium in muscle corresponds directly to the distribution of integrins (associated with the presence of satellite cells at the surface of myofibers) and the distribution of myonuclei, subsarcolemmal mitochondria, and myosin inside myofibers. We concluded that the perimysium-myofiber relationship reflects the existence of a mechanosensor system explaining short and long term muscle adaptations. Our investigations were essential for detecting the way of myofibers adaptations

Gli stili APA, Harvard, Vancouver, ISO e altri

6

Delaforge, Elise. "Dynamique structurale et fonctionnelle du domaine C-terminal de la protéine PB2 du virus de la grippe A". Thesis, Université Grenoble Alpes (ComUE), 2015. http://www.theses.fr/2015GREAV037/document.

Testo completo

Abstract (sommario):

La capacité du virus de la grippe aviaire à traverser la barrière des espèces et à devenir fortement pathogène chez les mammifères est un problème majeur de santé publique. Chez les oiseaux, la réplication a lieu dans l'intestin, à 4C, tandis que chez les humains elle a lieu dans l'appareil respiratoire, plus froid, à 33C. Il a été montré que l'adaptation à la température du virus de la grippe a lieu par de nombreuses mutations de la polymérase virale, notamment dans le domaine 627-NLS situé en C-terminal de la protéine PB2. Ce domaine est impliqué dans l'adaptation à l'hôte et interagit avec la protéine de l'hôte, importine alpha, étant donc indispensable pour l'entrée de la polymérase virale dans le noyau de la cellule [Tarendeau et al., 2008]. Les structures cristallographiques du 627-NLS et du complexe importine alpha/NLS existent. Cependant, lors de la superposition de ces structures via leur domaine NLS commun, un important choc stérique entre le domaine 627 et l'importine alpha devient évident. Ceci indique qu'une autre conformation du 627-NLS est requise pour l'interaction avec l'importine alpha [Boivin and Hart, 2011]. Dans cette étude, nous avons examiné les bases moléculaires de l'adaptation inter-espèces du virus à travers l'étude de la structure et de la dynamique du 627-NLS aviaire et humain. Nous avons identifié deux conformations du 627-NLS en échange lent (10-100 s-1), correspondant apparemment à une conformation ouverte et une conformation fermée des deux domaines. Nous proposons que la conformation ouverte du 627-NLS est la seule conformation compatible avec l'interaction avec l'importine alpha, et que l'équilibre entre conformation ouverte et fermée pourrait jouer le rôle de thermostat moléculaire, contrôlant l'efficacité de la réplication virale chez différents hôtes. La cinétique et la dynamique de ce comportement conformationnel important ainsi que de l'interaction entre le 627-NLS et l'importine alpha ont été caractérisées par résonance magnétique nucléaire (déplacements chimique, augmentation paramagnétique de la relaxation, relaxation de spin, transfert de saturation par l'échange chimique), combinée à la diffusion des rayons X et des neutrons aux petits angles ainsi qu'au transfert d'énergie par résonance de type Förster. Aussi, nous avons déterminé les affinités d'une série de mutants évolutifs du 627-NLS pour l'importine alpha et du 627-NLS aviaire ou humain pour différents isoformes de l'importine alpha, montrant que les affinités observées sont cohérentes avec les préférences d'interactions vues in vivo
The ability of avian influenza viruses to cross the species barrier and become dangerously pathogenic to mammalian hosts represents a major threat for human health. In birds the viral replication is carried out in the intestine at 40°C, while in humans it occurs in the cooler respiratory tract at 33°C. It has been shown that temperature adaption of the influenza virus occurs through numerous mutations in the viral polymerase, in particular in the C-terminal domain 627-NLS of the PB2 protein. This domain has already been shown to participate in host adaptation and is involved in importin alpha binding and therefore is required for entry of the viral polymerase into the nucleus [Tarendeau et al., 2008]. Crystallographic structures are available for 627-NLS and the complex importin alpha/NLS, however, a steric clash between importin alpha and the 627 domain becomes apparent when superimposing the NLS domain of the two structures, indicating that another conformation of 627-NLS is required for binding to importin alpha [Boivin and Hart, 2011]. Here we investigate the molecular basis of inter-species adaptation by studying the structure and dynamics of human and avian 627-NLS. We have identified two conformations of 627-NLS in slow exchange (10-100 s-1), corresponding to an apparently open and closed conformation of the two domains. We show that the equilibrium between closed and open conformations is strongly temperature dependent. We propose that the open conformation of 627-NLS is the only conformation compatible with binding to importin alpha and that the equilibrium between closed and open conformations may play a role as a molecular thermostat, controlling the efficiency of viral replication in the different species. The kinetics and domain dynamics of this important conformational behaviour and of the interaction between 627-NLS and importin alpha have been characterized using nuclear magnetic resonance chemical shifts, paramagnetic relaxation enhancement, spin relaxation and chemical exchange saturation transfer, in combination with X-ray and neutron small angle scattering and Förster resonance energy transfer. Also, we have determined the affinities of various evolutionnary mutants of 627-NLS to importin alpha and of avian and human 627-NLS to different isoforms of importin alpha, showing that the observed affinities are coherent with the preferred interactions seen in vivo

Gli stili APA, Harvard, Vancouver, ISO e altri

7

Lopez, Rémy. "Adaptation des méthodes “statistical energy analysis” (sea) aux problèmes d'électromagnétisme en cavités". Toulouse 3, 2006. http://www.theses.fr/2006TOU30045.

Testo completo

Abstract (sommario):

Modéliser des phénomènes électromagnétiques par des méthodes déterministes requiert une division du volume étudié en éléments discrets dont la taille est de l'ordre du dixième de la longueur d'onde. La demande en ressource informatique augmente donc avec la fréquence. De plus, compte tenu de la complexité des problèmes et des incertitudes sur les données d'entrées, il devient illusoire de réaliser un calcul déterministe pour chaque variable analysée. De nouvelles méthodes, dites énergétiques, sont développées pour étudier les systèmes grands devant la longueur d'onde. Elles permettent d'estimer statistiquement la valeur du champ à l'intérieur d'un système. Une de ces techniques, la Statistical Energy Analysis (SEA), développée en acoustique, est transposée ici en électromagnétisme. La SEA permet de décrire les échanges d'énergies entre les différents systèmes composant une structure. L'énergie de chaque système dépend des notions de mode de résonance, de perte et de couplage. Les paramètres liés à ces notions sont évalués analytiquement et numériquement. Une méthode de sous structuration automatique est également présentée Les résultats obtenus semblent confirmer l'intérêt de cette méthode
Modeling electromagnetic phenomena by deterministic methods requires a subdivision of the volume under study into a number of discrete elements with sizes of the order of tenth of the wavelength. So, the demand for computer resources significantly grows with increasing frequencies. Moreover, taking into account the complexity of the problems and the uncertainties on the input data, it becomes illusory to make a deterministic calculation for each studied variable. New methods, called energetic methods, were developed to study systems large in front of the wavelength. They allow to estimate statistically the value of the field inside a system One of these methods, the Statistical Energy Analysis (SEA), developed in acoustic, is transposed here in electromagnetism. The SEA allows to describe the exchanges of energy between the different systems of a structure. The energy of each system depends on the concepts of mode of resonance, loss and coupling. The parameters linked with these concepts are assessed by analytical formulae and numerical simulations. An automatic sub structuring method is also presented. The results obtained seem to confirm the interest of this method

Gli stili APA, Harvard, Vancouver, ISO e altri

8

Sidibe, Mamadou Gouro. "Métrologie des services audiovisuels dans un contexte multi-opérateurs et multi-domaines réseaux". Versailles-St Quentin en Yvelines, 2010. http://www.theses.fr/2010VERS0068.

Testo completo

Abstract (sommario):

L’accès aux services multimédia via des réseaux et terminaux hétérogènes est en constante augmentation, alors que la garantie de la qualité de service (QoS) de bout-en-bout reste un défi. Relever ce défi nécessite de déployer de nouvelles architectures de gestion incluant des agents surveillant les paramètres réseaux (NQoS), ainsi que la Qualité d’Expérience (QoE) de l'utilisateur. Nous proposons un Contrôleur Intégré de Gestion de QoS permettant un efficace approvisionnement, surveillance et adaptation des services vidéo basés sur la norme MPEG-21. Ensuite, nous proposons une solution de métrologie novatrice qui, prenant en charge la QoE, permet de surveiller dans les réseaux hétérogènes la connectivité des services à grande échelle et la qualité perçue par l'utilisateur. La solution définit une architecture comprenant des agents de métrologie pour les niveaux: nœud de réseau, réseau, applicatif et service. La solution définit également des protocoles de signalisation
Access to multimedia services over heterogeneous networks and terminals is of increasing market interest, while providing end-to-end (E2E) Quality of Service (QoS) guarantees is still a challenge. Solving this issue requires to deploy new E2E management architectures including components that monitor the network QoS (NQoS) parameters, as well as the Quality of Experience (QoE) of the user. In this thesis, we first propose an E2E Integrated QoS Management Supervisor for an efficient provisioning, monitoring and adaptation of video services using the MPEG-21 standard. We then propose a novel QoE-aware monitoring solution for large-scale service connectivity and user-perceived quality monitoring over heterogeneous networks. The solution specifies a scalable cross-layer monitoring architecture, comprising four types of QoS monitoring agents operating at node, network, application and service levels. It also specifies related intra/inter-domain signalling protocols

Gli stili APA, Harvard, Vancouver, ISO e altri

9

Rouquet, Géraldine. "Etude du rôle de l'opéron métabolique frz dans la virulence d'escherichia coli et dans son adaptation aux conditions environnementales". Thesis, Tours, 2010. http://www.theses.fr/2010TOUR4008.

Testo completo

Abstract (sommario):

L’opéron métabolique frz code les sous-unités d’un transporteur PTS de la sous-famille du fructose, un activateur transcriptionnel des systèmes PTS de la famille MgA (FrzR), deux cétoses-1,6-bisphosphate aldolases de type II, une kinase spécifique des sucres (famille ROK) et une protéine de la superfamille des cupines. Il est fortement associé aux souches d’Escherichia coli à virulence extra-intestinale. Nous avons montré qu’il procure un avantage aux bactéries lors de conditions de stress (peu d’oxygène, phase stationnaire de croissance, croissance dans le sérum et l’intestin) et qu’il est impliqué dans l’adhérence et l’internalisation de la bactérie dans diverses cellules eucaryotes, en régulant l’expression des fimbriae de type 1. L’activateur (FrzR) est impliqué dans ces phénotypes. A l’aide de microarrays, une série de gènes sous la dépendance du système Frz ont été identifiés. Nos données suggèrent que frz code un senseur de l’environnement permettant à E. coli de s’adapter à un environnement fluctuant en régulant notamment certains gènes de virulence et d’adaptation à l’hôte. Un modèle de régulation est présenté
The metabolic frz operon codes for three subunits of a PTS transporter of the fructose sub-family, for a transcriptional activator of PTS systems of the MgA family (FrzR), for two type II ketose-1,6-bisphosphate aldolases, for a sugar specific kinase (ROK family) and for a protein of the cupin superfamily. It is highly associated with Extra-intestinal Pathogenic Escherichia coli strains. We proved that frz promotes bacterial fitness under stressful conditions, (such as oxygen restriction, late stationary phase of growth or growth in serum or in the intestinal tract). Furthermore, we showed that frz is involved in adherence to and internalization of E. coli in several eukaryotic cells by regulating the expression of type 1 fimbriae. The FrzR activator is involved in these phenotypes. Microarrays, experiments allowed the identification of several genes under the dependence of the frz system. Our data suggest that frz codes for a sensor of the environment allowing E. coli to adapt to a fluctuating environment by regulating some virulence and host adaptation genes. A regulation model is presented

Gli stili APA, Harvard, Vancouver, ISO e altri

10

Alqasir, Hiba. "Apprentissage profond pour l'analyse de scènes de remontées mécaniques : amélioration de la généralisation dans un contexte multi-domaines". Thesis, Lyon, 2020. http://www.theses.fr/2020LYSES045.

Testo completo

Abstract (sommario):

Nous présentons notre travail sur la sécurité des télésièges par des techniques d'apprentissage profond dans le cadre du projet Mivao, qui vise à développer un système de vision par ordinateur qui acquiert des images de la station d'embarquement du télésiège, analyse les éléments essentiels et détecte les situations dangereuses. Dans ce scénario, nous avons différents télésièges répartis sur différentes stations de ski, avec une grande diversité de conditions d'acquisition et de géométries . Lorsque le système est installé pour un nouveau télésiège, l'objectif est d'effectuer une analyse de scène précise et fiable, étant donné le manque de données labellisées sur ce télésiège.Dans ce contexte, nous nous concentrons principalement sur le garde-corps du télésiège et proposons de classer chaque image en deux catégories, selon que le garde-corps est fermé ou ouvert. Il s'agit donc d'un problème de classification des images avec trois spécificités : (i) la catégorie d'image dépend d'un petit détail dans un fond encombré, (ii) les annotations manuelles ne sont pas faciles à obtenir, (iii) un classificateur formé sur certains télésièges devrait donner de bons résultats sur un nouveau. Pour guider le classificateur vers les zones importantes des images, nous avons proposé deux solutions : la détection d'objets et les réseaux siamois.Nos solutions sont motivées par la nécessité de minimiser les efforts d'annotation humaine tout en améliorant la précision du problème de la sécurité des télésièges. Cependant, ces contributions ne sont pas nécessairement limitées à ce contexte spécifique, et elles peuvent être appliquées à d'autres problèmes dans un contexte multi-domaine
This thesis presents our work on chairlift safety using deep learning techniques as part of the Mivao project, which aims to develop a computer vision system that acquires images of the chairlift boarding station, analyzes the crucial elements, and detects dangerous situations. In this scenario, we have different chairlifts spread over different ski resorts, with a high diversity of acquisition conditions and geometries; thus, each chairlift is considered a domain. When the system is installed for a new chairlift, the objective is to perform an accurate and reliable scene analysis, given the lack of labeled data on this new domain (chairlift).In this context, we mainly concentrate on the chairlift safety bar and propose to classify each image into two categories, depending on whether the safety bar is closed (safe) or open (unsafe). Thus, it is an image classification problem with three specific features: (i) the image category depends on a small detail (the safety bar) in a cluttered background, (ii) manual annotations are not easy to obtain, (iii) a classifier trained on some chairlifts should provide good results on a new one (generalization). To guide the classifier towards the important regions of the images, we have proposed two solutions: object detection and Siamese networks. Furthermore, we analyzed the generalization property of these two approaches. Our solutions are motivated by the need to minimize human annotation efforts while improving the accuracy of the chairlift safety problem. However, these contributions are not necessarily limited to this specific application context, and they may be applied to other problems in a multi-domain context

Gli stili APA, Harvard, Vancouver, ISO e altri

11

Ciobanu, Oana Alexandra. "Méthode de décomposition de domaine avec adaptation de maillage en espace-temps pour les équations d'Euler et de Navier-Stockes". Thesis, Paris 13, 2014. http://www.theses.fr/2014PA132052/document.

Testo completo

Abstract (sommario):

En mécanique des fluides, la simulation de phénomènes physiques de plus en plus complexes, en particulier instationnaires, nécessite des systèmes d’équations à nombre très élevé de degrés de liberté. Sous leurs formes originales, ces problèmes sont coûteux en temps CPU et ne permettent pas de faire une simulation sur une grande échelle de temps. Une formulation implicite, similaire à une méthode de Schwarz, avec une parallélisation simple par blocs et raccord explicite aux interfaces ne suffit plus à la résolution d’un tel système. Des méthodes de décomposition des domaines plus élaborées, adaptées aux nouvelles architectures, doivent être mises en place.Cette étude a consisté à élaborer un code de mécanique des fluides, parallèle, capable d’optimiser la convergence des méthodes du type Schwarz tout en améliorant la stabilité numérique et en diminuant le temps de calcul de la simulation. Une première partie a été l’étude de schémas numériques pour des problèmes stationnaires et instationnaires de type Euler et Navier–Stokes. Deuxièmement, une méthode de décomposition de domaine adaptive en espace-temps, a été proposée afin de profiter de l’échelle de temps caractéristique de la simulation dans chaque sous-domaine. Une troisième étude a été concentrée sur les moyens existants qui permettent de mettre en oeuvre ce code en parallèle (MPI, OPENMP, GPU). Des résultats numériques montrent l’efficacité de la méthode
Numerical simulations of more and more complex fluid dynamics phenomena, especially unsteady phenomena, require solving systems of equations with high degrees of freedom. Under their original form, these aerodynamic multi-scale problems are difficult to solve, costly in CPU time and do not allow simulations of large time scales. An implicit formulation, similar to the Schwarz method, with a simple block parallelisation and explicit coupling is no longer sufficient. More robust domain decomposition methods must be conceived so as to make use and adapt to the most of existent hardware.The main aim of this study was to build a parallel in space and in time CFD Finite Volumes code for steady/unsteady problems modelled by Euler and Navier-Stokes equations based on Schwarz method that improves consistency, accelerates convergence and decreases computational cost. First, a study of discretisation and numerical schemes to solve steady and unsteady Euler and Navier–Stokes problems has been conducted. Secondly, an adaptive timespace domain decomposition method has been proposed, as it allows local time stepping in each sub-domain. Thirdly, we have focused our study on the implementation of different parallel computing strategies (OpenMP, MPI, GPU). Numerical results illustrate the efficiency of the method

Gli stili APA, Harvard, Vancouver, ISO e altri

12

El, Boukkouri Hicham. "Domain adaptation of word embeddings through the exploitation of in-domain corpora and knowledge bases". Electronic Thesis or Diss., université Paris-Saclay, 2021. http://www.theses.fr/2021UPASG086.

Testo completo

Abstract (sommario):

Il existe, à la base de la plupart des systèmes de TAL, des représentations numériques appelées « plongements lexicaux » qui permettent à la machine de traiter, d'interagir avec et, dans une certaine mesure, de comprendre le langage humain. Ces plongements lexicaux nécessitent une quantité importante de textes afin d'être entraînés correctement, ce qui conduit souvent les praticiens du TAL à collecter et fusionner des textes provenant de sources multiples, mélangeant souvent différents styles et domaines (par exemple, des encyclopédies, des articles de presse, des articles scientifiques, etc.). Ces corpus dits du « domaine général » sont aujourd'hui la base sur laquelle s'entraînent la plupart des plongements lexicaux, limitant fortement leur utilisation dans des domaines plus spécifiques. En effet, les « domaines spécialisés » comme le domaine médical manifestent généralement assez de spécificités lexicales, sémantiques et stylistiques (par exemple, l'utilisation d'acronymes et de termes techniques) pour que les plongements lexicaux généraux ne soient pas en mesure de les représenter efficacement. Dans le cadre de cette thèse, nous explorons comment différents types de ressources peuvent être exploités afin soit d’entraîner de nouveaux plongements spécialisés, soit de spécialiser davantage des représentations préexistantes. Plus précisément, nous étudions d'abord comment des corpus de textes peuvent être utilisés à cette fin. En particulier, nous montrons que la taille du corpus ainsi que son degré de similarité au domaine d’intérêt jouent un rôle important dans ce processus puis proposons un moyen de tirer parti d'un petit corpus du domaine cible afin d’obtenir de meilleurs résultats dans des contextes à faibles ressources. Ensuite, nous abordons le cas des modèles de type BERT et observons que les vocabulaires généraux de ces modèles conviennent mal aux domaines spécialisés. Cependant, nous montrons des résultats indiquant que des modèles formés à l'aide de tels vocabulaires peuvent néanmoins être comparables à des systèmes entièrement spécialisés et utilisant des vocabulaires du domaine du domaine, ce qui nous amène à la conclusion que le ré-entraînement de modèles du domaine général est une approche tout à fait efficace pour construire des systèmes spécialisés. Nous proposons également CharacterBERT, une variante de BERT capable de produire des représentations de mots entiers en vocabulaire ouvert via la consultation des caractères de ces mots. Nous montrons des résultats indiquant que cette architecture conduit à une amélioration des performances dans le domaine médical tout en étant plus robuste aux fautes d'orthographe. Enfin, nous étudions comment des ressources externes sous forme de bases de connaissances et ontologies du domaine peuvent être exploitées pour spécialiser des représentations de mots préexistantes. Dans ce cadre, nous proposons une approche simple qui consiste à construire des représentations denses de bases de connaissances puis à combiner ces ``vecteurs de connaissances’’ avec les plongements lexicaux cibles. Nous généralisons cette approche et proposons également des Modules d'Injection de Connaissances, de petites couches neuronales permettant l'intégration de représentations de connaissances externes au sein des couches cachées de modèles à base de Transformers. Globalement, nous montrons que ces approches peuvent conduire à de meilleurs résultats, cependant, nous avons l'intuition que ces performances finales dépendent en fin de compte de la disponibilité de connaissances pertinentes pour la tâche cible au sein des bases de connaissances considérées. Dans l'ensemble, notre travail montre que les corpus et bases de connaissances du domaine peuvent être utilisés pour construire de meilleurs plongements lexicaux en domaine spécialisé. Enfin, afin de faciliter les recherches futures sur des sujets similaires, nous publions notre code et partageons autant que possible nos modèles pré-entraînés
There are, at the basis of most NLP systems, numerical representations that enable the machine to process, interact with and—to some extent—understand human language. These “word embeddings” come in different flavours but can be generally categorised into two distinct groups: on one hand, static embeddings that learn and assign a single definitive representation to each word; and on the other, contextual embeddings that instead learn to generate word representations on the fly, according to a current context. In both cases, training these models requires a large amount of texts. This often leads NLP practitioners to compile and merge texts from multiple sources, often mixing different styles and domains (e.g. encyclopaedias, news articles, scientific articles, etc.) in order to produce corpora that are sufficiently large for training good representations. These so-called “general domain” corpora are today the basis on which most word embeddings are trained, greatly limiting their use in more specific areas. In fact, “specialized domains” like the medical domain usually manifest enough lexical, semantic and stylistic idiosyncrasies (e.g. use of acronyms and technical terms) that general-purpose word embeddings are unable to effectively encode out-of-the-box. In this thesis, we explore how different kinds of resources may be leveraged to train domain-specific representations or further specialise preexisting ones. Specifically, we first investigate how in-domain corpora can be used for this purpose. In particular, we show that both corpus size and domain similarity play an important role in this process and propose a way to leverage a small corpus from the target domain to achieve improved results in low-resource settings. Then, we address the case of BERT-like models and observe that the general-domain vocabularies of these models may not be suited for specialized domains. However, we show evidence that models trained using such vocabularies can be on par with fully specialized systems using in-domain vocabularies—which leads us to accept re-training general domain models as an effective approach for constructing domain-specific systems. We also propose CharacterBERT, a variant of BERT that is able to produce word-level open-vocabulary representations by consulting a word's characters. We show evidence that this architecture leads to improved performance in the medical domain while being more robust to misspellings. Finally, we investigate how external resources in the form of knowledge bases may be leveraged to specialise existing representations. In this context, we propose a simple approach that consists in constructing dense representations of these knowledge bases then combining these knowledge vectors with the target word embeddings. We generalise this approach and propose Knowledge Injection Modules, small neural layers that incorporate external representations into the hidden states of a Transformer-based model. Overall, we show that these approaches can lead to improved results, however, we intuit that this final performance ultimately depends on whether the knowledge that is relevant to the target task is available in the input resource. All in all, our work shows evidence that both in-domain corpora and knowledge may be used to construct better word embeddings for specialized domains. In order to facilitate future research on similar topics, we open-source our code and share pre-trained models whenever appropriate

Gli stili APA, Harvard, Vancouver, ISO e altri

13

Sandu, Oana. "Domain adaptation for summarizing conversations". Thesis, University of British Columbia, 2011. http://hdl.handle.net/2429/33932.

Testo completo

Abstract (sommario):

The goal of summarization in natural language processing is to create abridged and informative versions of documents. A popular approach is supervised extractive summarization: given a training source corpus of documents with sentences labeled with their informativeness, train a model to select sentences from a target document and produce an extract. Conversational text is challenging to summarize because it is less formal, its structure depends on the modality or domain, and few annotated corpora exist. We use a labeled corpus of meeting transcripts as the source, and attempt to summarize a different target domain, threaded emails. We study two domain adaptation scenarios: a supervised scenario in which some labeled target domain data is available for training, and an unsupervised scenario with only unlabeled data in the target and labeled data available in a related but different domain. We implement several recent domain adaptation algorithms and perform a comparative study of their performance. We also compare the effectiveness of using a small set of conversation-specific features with a large set of raw lexical and syntactic features in domain adaptation. We report significant improvements of the algorithms over their baselines. Our results show that in the supervised case, given the amount of email data available and the set of features specific to conversations, training directly in-domain and ignoring the out-of-domain data is best. With only the more domain-specific lexical features, though overall performance is lower, domain adaptation can effectively leverage the lexical features to improve in both the supervised and unsupervised scenarios.

Gli stili APA, Harvard, Vancouver, ISO e altri

14

Htike, Kyaw Kyaw. "Domain adaptation for pedestrian detection". Thesis, University of Leeds, 2014. http://etheses.whiterose.ac.uk/7290/.

Testo completo

Abstract (sommario):

Object detection is an essential component of many computer vision systems. The increase in the amount of collected digital data and new applications of computer vision have generated a demand for object detectors for many different types of scenes digitally captured in diverse settings. The appearance of objects captured across these different scenarios can vary significantly, causing readily available state-of-the-art object detectors to perform poorly in many of the scenes. One solution is to annotate and collect labelled data for each new scene and train a scene-specific object detector that is specialised to perform well for that scene, but such a method is labour intensive and impractical. In this thesis, we propose three novel contributions to learn scene-specific pedestrian detectors for scenes with minimal human supervision effort. In the first and second contributions, we formulate the problem as unsupervised domain adaptation in which a readily available generic pedestrian detector is automatically adapted to specific scenes (without any labelled data from the scenes). In the third contribution, we formulate it as a weakly supervised learning algorithm requiring annotations of only pedestrian centres. The first contribution is a detector adaptation algorithm using joint dataset feature learning. We use state-of-the-art deep learning for the purpose of detector adaptation by exploiting the assumption that the data lies on a low dimensional manifold. The algorithm significantly outperforms a state-of-the-art approach that makes use of a similar manifold assumption. The second contribution presents an efficient detector adaptation algorithm that makes effective use of cues (e.g spatio-temporal constraints) available in video. We show that, for videos, such cues can dramatically help with the detector adaptation. We extensively compare our approach with state-of-the-art algorithms and show that our algorithm outperforms the competing approaches despite being simpler to implement and apply. In the third contribution, we approach the task of reducing manual annotation effort by formulating the problem as a weakly supervised learning algorithm that requires annotation of only approximate centres of pedestrians (instead of the usual precise bounding boxes). Instead of assuming the availability of a generic detector and adapting it to new scenes as in the first two contributions, we collect manual annotation for new scenes but make the annotation task easier and faster. Our algorithm reduces the amount of manual annotation effort by approximately four times while maintaining a similar detection performance as the standard training methods. We evaluate each of the proposed algorithms on two challenging publicly available video datasets.

Gli stili APA, Harvard, Vancouver, ISO e altri

15

Rahman, MD Hafizur. "Domain adaptation for speaker verification". Thesis, Queensland University of Technology, 2018. https://eprints.qut.edu.au/116511/1/MD%20Hafizur_Rahman_Thesis.pdf.

Testo completo

Abstract (sommario):

This PhD research developed new approaches to address speaker recognition system development using limited development data. Investigations in this program focused on finding the minimum in-domain data requirements for an i-vector PLDA system, leading to unsupervised and supervised methods for compensating domain mismatch from the training data to improve the current state of the art i-vector PLDA system speaker verification system.

Gli stili APA, Harvard, Vancouver, ISO e altri

16

Rahman, Mohammad Mahfujur. "Deep domain adaptation and generalisation". Thesis, Queensland University of Technology, 2020. https://eprints.qut.edu.au/205619/1/Mohammad%20Mahfujur_Rahman_Thesis.pdf.

Testo completo

Abstract (sommario):

This thesis addresses a critical problem in computer vision of dealing with dataset bias between source and target environments. Variations in image data can arise from multiple factors including contrasts in picture quality (shading, brightness, colour, resolution, and occlusion), diverse backgrounds, distinct circumstances, changes in camera viewpoint, and implicit heterogeneity of the samples themselves. This research developed strategies to address this domain shift problem for the object recognition task. Several domain adaptation and generalization approaches based on deep neural networks were introduced to improve poor performance due to domain shift or domain bias.

Gli stili APA, Harvard, Vancouver, ISO e altri

17

Rubino, Raphaël. "Traduction automatique statistique et adaptation à un domaine spécialisé". Phd thesis, Université d'Avignon, 2011. http://tel.archives-ouvertes.fr/tel-00879945.

Testo completo

Abstract (sommario):

Nous avons observé depuis plusieurs années l'émergence des approches statistiques pour la traduction automatique. Cependant, l'efficacité des modèles construits est soumise aux variabilités inhérentes au langage naturel. Des études ont montré la présence de vocabulaires spécifique et général composant les corpus de textes de domaines spécialisés. Cette particularité peut être prise en charge par des ressources terminologiques comme les lexiques bilingues.Toutefois, nous pensons que si le vocabulaire est différent entre des textes spécialisés ou génériques, le contenu sémantique et la structure syntaxique peuvent aussi varier. Dans nos travaux,nous considérons la tâche d'adaptation aux domaines spécialisés pour la traduction automatique statistique selon deux axes majeurs : l'acquisition de lexiques bilingues et l'édition a posteriori de traductions issues de systèmes automatiques. Nous évaluons l'efficacité des approches proposées dans un contexte spécialisé : le domaine médical. Nos résultats sont comparés aux travaux précédents concernant cette tâche. De manière générale, la qualité des traductions issues de systèmes automatiques pour le domaine médical est améliorée par nos propositions. Des évaluations en oracle tendent à montrer qu'il existe une marge de progression importante

Gli stili APA, Harvard, Vancouver, ISO e altri

18

Cardace, Adriano. "Learning Features Across Tasks and Domains". Master's thesis, Alma Mater Studiorum - Università di Bologna, 2020. http://amslaurea.unibo.it/20050/.

Testo completo

Abstract (sommario):

The absence of in-domain labeled data hinders the applicability of powerful deep neural networks. Unsupervised Domain Adaptation (UDA) methods have emerged to exploit such models even when labeled data is not available in the target domain. All these techniques aim to reduce the distribution shift problem that afflicts these models when trained on one dataset and tested in a different one. However, most of the works, do not consider relationships among tasks to further boost performances. In this thesis, we study a recent method called AT/DT (Across Tasks Domain Transfer), that seeks to apply Domain Adaptation together with Task Adaptation, leveraging on the correlation of two popular Vision tasks such as Semantic Segmentation and Monocular Depth Estimation. Inspired by the Domain Adaptation literature, we propose many extensions to the original work and show how these enhance the framework performances. Our contributions are applied at different levels: we first study how different architectures affect the transferability of features across tasks. We further improve performances by deploying Adversarial training. Finally, we explore the possibility of replacing Depth Estimation with popular Self-supervised tasks, demonstrating that two tasks must be semantically connected to be able to transfer features among them.

Gli stili APA, Harvard, Vancouver, ISO e altri

19

Saporta, Antoine. "Domain Adaptation for Urban Scene Segmentation". Electronic Thesis or Diss., Sorbonne université, 2022. http://www.theses.fr/2022SORUS115.

Testo completo

Abstract (sommario):

Cette thèse attaque certains des verrous scientifiques des systèmes de perception à base de réseaux de neurones des véhicules autonomes. Une classe d'outils abordée dans cette thèse pour limiter les besoins de données étiquetées est celle de l'adaptation de domaine. Celle-ci permet la généralisation à des données dites cibles qui partagent des structures avec les données annotées dites sources permettant la supervision mais qui suivent néanmoins une distribution statistique différente. D'abord, nous étudions l'introduction d'information privilégiée dans les données sources, par exemple des annotations de profondeur. La stratégie proposée BerMuDA appuie son adaptation de domaine sur une représentation multimodale obtenue par fusion bilinéaire, modélisant des interactions complexes entre segmentation et profondeur. Ensuite, nous examinons les stratégies d'auto-apprentissage en adaptation de domaine, reposant sur la sélection de prédictions sur les données cibles non étiquetées, servant de pseudo-étiquettes. Nous proposons deux nouveaux critères de sélection: d'abord, un critère entropique avec ESL; puis, avec ConDA, utilisant une estimation de la probabilité de la vraie classe. Enfin, l'extension des scénarios d'adaptation à plusieurs domaines cibles ainsi que dans un cadre d'apprentissage continu est proposée. Deux approches sont présentées pour étendre les méthodes adversaires traditionnelles à l'adaptation de domaine multi-cible: Multi-Dis. et MTKT. Dans un cadre d'apprentissage continu, les domaines cibles sont découverts séquentiellement et sans répétition. L'approche proposée CTKT adapte MTKT à ce nouveau problème pour lutter contre l'oubli catastrophique
This thesis tackles some of the scientific locks of perception systems based on neural networks for autonomous vehicles. This dissertation discusses domain adaptation, a class of tools aiming at minimizing the need for labeled data. Domain adaptation allows generalization to so-called target data that share structures with the labeled so-called source data allowing supervision but nevertheless following a different statistical distribution. First, we study the introduction of privileged information in the source data, for instance, depth labels. The proposed strategy, BerMuDA, bases its domain adaptation on a multimodal representation obtained by bilinear fusion, modeling complex interactions between segmentation and depth. Next, we examine self-supervised learning strategies in domain adaptation, relying on selecting predictions on the unlabeled target data, serving as pseudo-labels. We propose two new selection criteria: first, an entropic criterion with ESL; then, with ConDA, using an estimate of the true class probability. Finally, the extension of adaptation scenarios to several target domains as well as in a continual learning framework is proposed. Two approaches are presented to extend traditional adversarial methods to multi-target domain adaptation: Multi-Dis. and MTKT. In a continual learning setting for which the target domains are discovered sequentially and without rehearsal, the proposed CTKT approach adapts MTKT to this new problem to tackle catastrophic forgetting

Gli stili APA, Harvard, Vancouver, ISO e altri

20

Xu, Jiaolong. "Domain adaptation of deformable part-based models". Doctoral thesis, Universitat Autònoma de Barcelona, 2015. http://hdl.handle.net/10803/290266.

Testo completo

Abstract (sommario):

La detecció de vianants és crucial per als sistemes d’assistència a la conducció (ADAS). Disposar d’un classificador precís és fonamental per a un detector de vianants basat en visió. Al entrenar un classificador, s’assumeix que les característiques de les dades d’entrenament segueixen la mateixa distribució de probabilitat que la de les dades de prova. Tot i això, a la pràctica, aquesta assumpció pot no complir-se per diferents causes. En aquests casos, en la comunitat de visió per computador és cada cop més comú utilitzar tècniques que permeten adaptar els classificadors existents del seu entorn d’entrenament (domini d’origen) al nou entorn de prova (domini de destí). En aquesta tesi ens centrem en l’adaptació de domini dels detectors de vianants basats en models deformables basats en parts (DPMs). Com a prova de concepte, utilitzem dades sintètiques com a domini d’origen (món virtual) i adaptem el detector DPM entrenat en el món virtual per a funcionar en diferents escenaris reals. Començem explotant al màxim les capacitats de detecció del DPM entrenant en dades del món virtual, però, tot i això, al aplicar-lo a diferents conjunts del món real, el detector encara perd poder de discriminació degut a les diferències entre el món virtual i el real. És per això, que ens centrem en l’adaptació de domini del DPM. Per començar, considerem un únic domini d’origen per a adaptar-lo a un únic domini de destí mitjançant dos mètodes d’aprenentatge per lots, l’A-SSVM i el SASSVM. Després, l’ampliem a treballar amb múltiples (sub-)dominis mitjançant una adaptació progressiva, utilitzant una jerarquia adaptativa basada en SSVM (HASSVM) en el procés d’optimització. Finalment, extenem HA-SSVM per a aconseguir un detector que s’adapti de forma progressiva i sense intervenció humana al domini de destí. Cal destacar que cap dels mètodes proposats en aquesta tesi requereix visitar les dades del domini d’origen. L’evaluació dels resultats, realitzada amb el sistema d’evaluació de Caltech, mostra que el SA-SSVM millora lleugerament respecte el ASSVM i millora en 15 punts respecte el detector no adaptat. El model jeràrquic entrenat mitjançant el HA-SSVM encara millora més els resultats de la adaptació de domini. Finalment, el mètode sequencial d’adaptació de domini ha demostrat que pot obtenir resultats comparables a la adaptació per lots, però sense necessitat d’etiquetar manualment cap exemple del domini de destí. L’adaptació de domini aplicada a la detecció de vianants és de gran importància i és una àrea que es troba relativament sense explorar. Desitgem que aquesta tesi pugui assentar les bases del treball futur d’aquesta àrea.
La detección de peatones es crucial para los sistemas de asistencia a la conducción (ADAS). Disponer de un clasificador preciso es fundamental para un detector de peatones basado en visión. Al entrenar un clasificador, se asume que las características de los datos de entrenamiento siguen la misma distribución de probabilidad que las de los datos de prueba. Sin embargo, en la práctica, esta asunción puede no cumplirse debido a diferentes causas. En estos casos, en la comunidad de visión por computador cada vez es más común utilizar técnicas que permiten adaptar los clasificadores existentes de su entorno de entrenamiento (dominio de origen) al nuevo entorno de prueba (dominio de destino). En esta tesis nos centramos en la adaptación de dominio de los detectores de peatones basados en modelos deformables basados en partes (DPMs). Como prueba de concepto, usamos como dominio de origen datos sintéticos (mundo virtual) y adaptamos el detector DPM entrenado en el mundo virtual para funcionar en diferentes escenarios reales. Comenzamos explotando al máximo las capacidades de detección del DPM entrenado en datos del mundo virtual pero, aun así, al aplicarlo a diferentes conjuntos del mundo real, el detector todavía pierde poder de discriminaci ón debido a las diferencias entre el mundo virtual y el real. Es por ello que nos centramos en la adaptación de dominio del DPM. Para comenzar, consideramos un único dominio de origen para adaptarlo a un único dominio de destino mediante dos métodos de aprendizaje por lotes, el A-SSVM y SA-SSVM. Después, lo ampliamos a trabajar con múltiples (sub-)dominios mediante una adaptación progresiva usando una jerarquía adaptativa basada en SSVM (HA-SSVM) en el proceso de optimización. Finalmente, extendimos HA-SSVM para conseguir un detector que se adapte de forma progresiva y sin intervención humana al dominio de destino. Cabe destacar que ninguno de los métodos propuestos en esta tesis requieren visitar los datos del dominio de origen. La evaluación de los resultados, realizadas con el sistema de evaluación de Caltech, muestran que el SA-SSVM mejora ligeramente respecto al A-SSVM y mejora en 15 puntos respecto al detector no adaptado. El modelo jerárquico entrenado mediante el HA-SSVM todavía mejora más los resultados de la adaptación de dominio. Finalmente, el método secuencial de adaptación de domino ha demostrado que puede obtener resultados comparables a la adaptación por lotes pero sin necesidad de etiquetar manualmente ningún ejemplo del dominio de destino. La adaptación de domino aplicada a la detección de peatones es de gran importancia y es un área que se encuentra relativamente sin explorar. Deseamos que esta tesis pueda sentar las bases del trabajo futuro en esta área.
On-board pedestrian detection is crucial for Advanced Driver Assistance Systems (ADAS). An accurate classi cation is fundamental for vision-based pedestrian detection. The underlying assumption for learning classi ers is that the training set and the deployment environment (testing) follow the same probability distribution regarding the features used by the classi ers. However, in practice, there are di erent reasons that can break this constancy assumption. Accordingly, reusing existing classi ers by adapting them from the previous training environment (source domain) to the new testing one (target domain) is an approach with increasing acceptance in the computer vision community. In this thesis we focus on the domain adaptation of deformable part-based models (DPMs) for pedestrian detection. As a prof of concept, we use a computer graphic based synthetic dataset, i.e. a virtual world, as the source domain, and adapt the virtual-world trained DPM detector to various real-world dataset. We start by exploiting the maximum detection accuracy of the virtual-world trained DPM. Even though, when operating in various real-world datasets, the virtualworld trained detector still su er from accuracy degradation due to the domain gap of virtual and real worlds. We then focus on domain adaptation of DPM. At the rst step, we consider single source and single target domain adaptation and propose two batch learning methods, namely A-SSVM and SA-SSVM. Later, we further consider leveraging multiple target (sub-)domains for progressive domain adaptation and propose a hierarchical adaptive structured SVM (HA-SSVM) for optimization. Finally, we extend HA-SSVM for the challenging online domain adaptation problem, aiming at making the detector to automatically adapt to the target domain online, without any human intervention. All of the proposed methods in this thesis do not require revisiting source domain data. The evaluations are done on the Caltech pedestrian detection benchmark. Results show that SA-SSVM slightly outperforms A-SSVM and avoids accuracy drops as high as 15 points when comparing with a non-adapted detector. The hierarchical model learned by HA-SSVM further boosts the domain adaptation performance. Finally, the online domain adaptation method has demonstrated that it can achieve comparable accuracy to the batch learned models while not requiring manually label target domain examples. Domain adaptation for pedestrian detection is of paramount importance and a relatively unexplored area. We humbly hope the work in this thesis could provide foundations for future work in this area.

Gli stili APA, Harvard, Vancouver, ISO e altri

21

Shahabuddin, Sharmeen. "Compressed Domain Spatial Adaptation of H264 Videos". Thesis, University of Ottawa (Canada), 2010. http://hdl.handle.net/10393/28787.

Testo completo

Abstract (sommario):

A gigantic amount of multimedia contents is readily available for an ever growing consumer base at present due to advances in video coding technology and standardization along with rapid improvements of storage capacity and computing power. However, today's pervasive media environment which includes heterogeneous terminals and networks presents a serious obstacle in achieving seamless access to these contents. Storing individual content in several formats taking into account a wide variety of possible user preferences and resource constraints or adapting the content on the fly by cascaded decoding/re-encoding is not at all suitable for real-time applications. In this thesis, an innovative spatial adaptation technique for H.264/AVC videos is presented that adapts videos in the compressed domain without decoding/re-encoding them. The proposed adaptation strategy encodes each video frame into several slices according to the proposed slicing strategies. The adaptation process can reduce video size by cropping individual frames by dropping the undesired slices prior to transmitting that video to the clients. In order to ensure codec independence, structured metadata-based adaptation utilizing MPEG21 generic Bitstream Syntax Description (gBSD) is used to perform adaptation operations. The adaptation process is completely transparent to the client as the adapted video-stream is H.264 standard compliant. A prototype is developed for the proposed scheme. The performance results of the prototype are presented in this thesis. The results show that the proposed spatial adaptation scheme is very suitable and effective for adaptation of both pre-recorded and live video streams.

Gli stili APA, Harvard, Vancouver, ISO e altri

22

Herndon, Nic. "Domain adaptation algorithms for biological sequence classification". Diss., Kansas State University, 2016. http://hdl.handle.net/2097/35242.

Testo completo

Abstract (sommario):

Doctor of Philosophy
Department of Computing and Information Sciences
Doina Caragea
The large volume of data generated in the recent years has created opportunities for discoveries in various fields. In biology, next generation sequencing technologies determine faster and cheaper the exact order of nucleotides present within a DNA or RNA fragment. This large volume of data requires the use of automated tools to extract information and generate knowledge. Machine learning classification algorithms provide an automated means to annotate data but require some of these data to be manually labeled by human experts, a process that is costly and time consuming. An alternative to labeling data is to use existing labeled data from a related domain, the source domain, if any such data is available, to train a classifier for the domain of interest, the target domain. However, the classification accuracy usually decreases for the domain of interest as the distance between the source and target domains increases. Another alternative is to label some data and complement it with abundant unlabeled data from the same domain, and train a semi-supervised classifier, although the unlabeled data can mislead such classifier. In this work another alternative is considered, domain adaptation, in which the goal is to train an accurate classifier for a domain with limited labeled data and abundant unlabeled data, the target domain, by leveraging labeled data from a related domain, the source domain. Several domain adaptation classifiers are proposed, derived from a supervised discriminative classifier (logistic regression) or a supervised generative classifier (naïve Bayes), and some of the factors that influence their accuracy are studied: features, data used from the source domain, how to incorporate the unlabeled data, and how to combine all available data. The proposed approaches were evaluated on two biological problems -- protein localization and ab initio splice site prediction. The former is motivated by the fact that predicting where a protein is localized provides an indication for its function, whereas the latter is an essential step in gene prediction.

Gli stili APA, Harvard, Vancouver, ISO e altri

23

Shu, Le. "Graph and Subspace Learning for Domain Adaptation". Diss., Temple University Libraries, 2015. http://cdm16002.contentdm.oclc.org/cdm/ref/collection/p245801coll10/id/363757.

Testo completo

Abstract (sommario):

Computer and Information Science
Ph.D.
In many practical problems, given that the instances in the training and test may be drawn from different distributions, traditional supervised learning can not achieve good performance on the new domain. Domain adaptation algorithms are therefore designed to bridge the distribution gap between training (source) data and test (target) data. In this thesis, I propose two graph learning and two subspace learning methods for domain adaptation. Graph learning methods use a graph to model pairwise relations between instances and then minimize the domain discrepancy based on the graphs directly. The first effort we make is to propose a novel locality preserving projection method for domain adaptation task, which can find a linear mapping preserving the intrinsic structure for both source and target domains. We first construct two graphs encoding the neighborhood information for source and target domains separately. We then find linear projection coefficients which have the property of locality preserving for each graph. Instead of combing the two objective terms under compatibility assumption and requiring the user to decide the importance of each objective function, we propose a multi-objective formulation for this problem and solve it simultaneously using Pareto optimization. Pareto optimization allows multiple objectives to compete with each other in deciding the optimal trade-off. We use generalized eigen-decomposition to find the pareto frontier, which captures all possible good linear projection coefficients that are preferred by one or more objectives. The second effort is to directly improve the pair-wise similarities between instances in the same domain as well as in different domains. We propose a novel method to solve domain adaptation task in a transductive setting. The proposed method bridges the distribution gap between source domain and target domain through affinity learning. It exploits the existence of a subset of data points in target domain which distribute similarly to the data points in the source domain. These data points act as the bridge that facilitates the data similarities propagation across domains. We also propose to control the relative importance of intra- and inter- domain similarities to boost the similarity propagation. In our approach, we first construct the similarity matrix which encodes both the intra- and inter- domain similarities. We then learn the true similarities among data points in joint manifold using graph diffusion. We demonstrate that with improved similarities between source and target data, spectral embedding provides a better data representation, which boosts the prediction accuracy. Subspace learning methods aim to find a new coordinate system, in which the domain discrepancy is minimized. In this thesis, we refer to subspace-based method as those which model the domain shift between two subspaces directly. Our first effort is to propose a novel linear subspace learning approach for domain adaptation. Our key observation is that in many real world problems, such as image classification with blurred test images or cross domain text classification, domain shift can be modeled by a linear transformation between the source and target domain (intrinsically linear transformation between two subspaces underlying the source and target data). Motivated by this observation, our method explicitly aligns the data in two domains using a linear transformation while simultaneously finding a subspace which preserves the most data variance. With explicit data alignment, the subspace learning is formulated as minimizing of a PCA-like objective, which consists of two variables: the basis vectors of the common subspace and the linear transformation between two domains. We show that the optimization can be solved efficiently using an iterative algorithm based on alternating minimization, and prove its convergence to a local optimum. Our method can also integrate the label information of source data, which further improves the robustness of the subspace learning and yields better prediction. Existing subspace based domain adaptation methods assume that data lie in a single low dimensional subspace. This assumption is too strong in many real world applications especially considering the domain could be a mixture of latent domains with significant inner-domain variations that should not be neglected. In our second approach, the key idea is to assume the data lie in a union of multiple low dimensional subspaces, which relaxes the common assumption above. We propose a novel two step subspace based domain adaptation algorithm: in subspaces discovery step, we cluster the source and target data using subspace clustering algorithm and estimate the subspace for each cluster using principal component analysis; in domain adaptation step, we propose a novel multiple subspace alignment (Multi-SA) algorithm, in which we identify one common subspace that aligns well with both source and target subspaces, and therefore, best preserves the variance for both domains. To solve this alignment problem jointly for multiple subspaces, we formulate this problem as solving an optimization problem that minimizes the weighted sum of multiple alignment costs. A higher weight is assigned to a source subspace if its label distribution has smaller distance, measured by KL divergence, compared to the overall label distribution. By putting more weights on those subspaces, the learned common subspace is able to to preserve the distinctive information.
Temple University--Theses

Gli stili APA, Harvard, Vancouver, ISO e altri

24

Di, Bella Laura. "Women's adaptation to STEM domains : generalised effects on judgement and cognition". Thesis, University of Kent, 2013. http://ethos.bl.uk/OrderDetails.do?uin=uk.bl.ethos.654096.

Testo completo

Abstract (sommario):

The chronic underrepresentation of women in STEM (Science, Technology, Engineering and Maths) fields is a recognised, and widely investigated, social issue. This thesis reports a programme of research testing whether women's experience in STEM can have a psychological impact that extends beyond their academic domain. Four studies examined the differential effects of counter-stereotypical experiences on women from STEM and non-STEM fields. Results provided only partial support to the hypothesis, with two studies detecting a differential effect of exposure to counter-stereotypical priming, and two studies detecting superior STEM women's performances regardless of priming condition. Further investigation is required to interpret more accurately both the broader impact of chronic exposure to challenging experiences, and also the interaction between such experiences and further counter-stereotypical priming. Hopefully, this will support the call for a novel perspective on the issue of promoting women's entry to STEM field; that is, exploring not only the barriers that keep women away from the sciences, but also the benefits associated with entering those fields. Four more studies investigated whether exposure to stereotyping not only reduces women's willingness to engage in STEM, but stifles broader egalitarian concerns. Only one study broadly supported the hypothesis, by showing that women exposed to gender-occupational stereotypes felt less angry about the condition of women in STEM, endorsed more the negative stereotypes about women in STEM, and were marginally more resistant to social change in general. This line of research has the potential to highlight the importance of tackling gender stereotypes not only because they exclude women from maledominated careers, but also because of a potentially pervasive negative impact on broader egalitarian concerns. By exploring the issue of women in STEM from novel perspectives, this thesis contributes to the public and scholarly debate of the impact of stereotyping and gender inequalities in STEM fields.

Gli stili APA, Harvard, Vancouver, ISO e altri

25

Selvaggi, Kevin. "Synthetic-to-Real Domain Adaptation for Autonomous Driving". Master's thesis, Alma Mater Studiorum - Università di Bologna, 2020.

Cerca il testo completo

Abstract (sommario):

Questa tesi rappresenta il risultato di un tirocinio svolto presso il reparto di test di Siemens Industry Software NV Leuven, in Belgio. Il primo obiettivo è stato quello di avere una visione generale sul settore della guida autonoma e delle relative tecnologie: si presenta quindi un'analisi della letteratura. Il campo è stato quindi ristretto ai riconoscitori di oggetti 2D che utilizzano sensori automotive come dispositivi di input. Dopo uno studio dello stato dell'arte di architetture di reti neurali e dei corrispondenti dataset usati per l'allenamento in questo settore, la domanda di ricerca è stata come validare la robustezza del sistema in condizioni reali, in particolare in degradate condizioni del manto stradale. Non essendo stato possibile portare a termine una campagna di test su strada per via dell'emergenza sanitaria, la decisione è stata di proseguire lo studio mediante la sperimentazione e testing virtuale. Questo inoltre ha permesso di estendere gli studi sulla scarsità di dati che affligge l'apprendimento automatico, consentendo di verificare se un ambiente virtuale può essere utile per la generazione di dati per l'allenamento e se può aiutare ad affrontare il problema del'adattamento del dominio. Quest'ultimo è stato affontato sotto due sfumature diverse. Da un lato l'intrinsica differenza tra dati reali e dati sintetici, dall'altro l'abilità di un sistema di adattarsi ad un nuovo dominio reale, che quindi presenta, ad esempio, delle condizioni ambientali differenti.

Gli stili APA, Harvard, Vancouver, ISO e altri

26

Sopova, Oleksandra. "Domain adaptation for classifying disaster-related Twitter data". Kansas State University, 2017. http://hdl.handle.net/2097/35388.

Testo completo

Abstract (sommario):

Master of Science
Department of Computing and Information Sciences
Doina Caragea
Machine learning is the subfield of Artificial intelligence that gives computers the ability to learn without being explicitly programmed, as it was defined by Arthur Samuel - the American pioneer in the field of computer gaming and artificial intelligence who was born in Emporia, Kansas. Supervised Machine Learning is focused on building predictive models given labeled training data. Data may come from a variety of sources, for instance, social media networks. In our research, we use Twitter data, specifically, user-generated tweets about disasters such as floods, hurricanes, terrorist attacks, etc., to build classifiers that could help disaster management teams identify useful information. A supervised classifier trained on data (training data) from a particular domain (i.e. disaster) is expected to give accurate predictions on unseen data (testing data) from the same domain, assuming that the training and test data have similar characteristics. Labeled data is not easily available for a current target disaster. However, labeled data from a prior source disaster is presumably available, and can be used to learn a supervised classifier for the target disaster. Unfortunately, the source disaster data and the target disaster data may not share the same characteristics, and the classifier learned from the source may not perform well on the target. Domain adaptation techniques, which use unlabeled target data in addition to labeled source data, can be used to address this problem. We study single-source and multi-source domain adaptation techniques, using Nave Bayes classifier. Experimental results on Twitter datasets corresponding to six disasters show that domain adaptation techniques improve the overall performance as compared to basic supervised learning classifiers. Domain adaptation is crucial for many machine learning applications, as it enables the use of unlabeled data in domains where labeled data is not available.

Gli stili APA, Harvard, Vancouver, ISO e altri

27

Thornström, Johan. "Domain Adaptation of Unreal Images for Image Classification". Thesis, Linköpings universitet, Datorseende, 2019. http://urn.kb.se/resolve?urn=urn:nbn:se:liu:diva-165758.

Testo completo

Abstract (sommario):

Deep learning has been intensively researched in computer vision tasks like im-age classification. Collecting and labeling images that these neural networks aretrained on is labor-intensive, which is why alternative methods of collecting im-ages are of interest. Virtual environments allow rendering images and automaticlabeling, which could speed up the process of generating training data and re-duce costs.This thesis studies the problem of transfer learning in image classificationwhen the classifier has been trained on rendered images using a game engine andtested on real images. The goal is to render images using a game engine to createa classifier that can separate images depicting people wearing civilian clothingor camouflage. The thesis also studies how domain adaptation techniques usinggenerative adversarial networks could be used to improve the performance ofthe classifier. Experiments show that it is possible to generate images that canbe used for training a classifier capable of separating the two classes. However,the experiments with domain adaptation were unsuccessful. It is instead recom-mended to improve the quality of the rendered images in terms of features usedin the target domain to achieve better results.

Gli stili APA, Harvard, Vancouver, ISO e altri

28

Shah, Darsh J. (Darsh Jaidip). "Multi-source domain adaptation with mixture of experts". Thesis, Massachusetts Institute of Technology, 2019. https://hdl.handle.net/1721.1/121741.

Testo completo

Abstract (sommario):

Thesis: S.M., Massachusetts Institute of Technology, Department of Electrical Engineering and Computer Science, 2019
Cataloged from PDF version of thesis.
Includes bibliographical references (pages 35-37).
We propose a mixture-of-experts approach for unsupervised domain adaptation from multiple sources. The key idea is to explicitly capture the relationship between a target example and different source domains. This relationship, expressed by a point-to-set metric, determines how to combine predictors trained on various domains. The metric is learned in an unsupervised fashion using meta-training. Experimental results on sentiment analysis and part-of-speech tagging demonstrate that our approach consistently outperforms multiple baselines and can robustly handle negative transfer.
by Darsh J. Shah.
S.M.
S.M. Massachusetts Institute of Technology, Department of Electrical Engineering and Computer Science

Gli stili APA, Harvard, Vancouver, ISO e altri

29

Yang, Baoyao. "Distribution alignment for unsupervised domain adaptation: cross-domain feature learning and synthesis". HKBU Institutional Repository, 2018. https://repository.hkbu.edu.hk/etd_oa/556.

Testo completo

Abstract (sommario):

In recent years, many machine learning algorithms have been developed and widely applied in various applications. However, most of them have considered the data distributions of the training and test datasets to be similar. This thesis concerns on the decrease of generalization ability in a test dataset when the data distribution is different from that of the training dataset. As labels may be unavailable in the test dataset in practical applications, we follow the effective approach of unsupervised domain adaptation and propose distribution alignment methods to improve the generalization ability of models learned from the training dataset in the test dataset. To solve the problem of joint distribution alignment without target labels, we propose a new criterion of domain-shared group sparsity that is an equivalent condition for equal conditional distribution. A domain-shared group-sparse dictionary learning model is built with the proposed criterion, and a cross-domain label propagation method is developed to learn a target-domain classifier using the domain-shared group-sparse representations and the target-specific information from the target data. Experimental results show that the proposed method achieves good performance on cross-domain face and object recognition. Moreover, most distribution alignment methods have not considered the difference in distribution structures, which results in insufficient alignment across domains. Therefore, a novel graph alignment method is proposed, which aligns both data representations and distribution structural information across the source and target domains. An adversarial network is developed for graph alignment by mapping both source and target data to a feature space where the data are distributed with unified structure criteria. Promising results have been obtained in the experiments on cross-dataset digit and object recognition. Problem of dataset bias also exists in human pose estimation across datasets with different image qualities. Thus, this thesis proposes to synthesize target body parts for cross-domain distribution alignment, to address the problem of cross-quality pose estimation. A translative dictionary is learned to associate the source and target domains, and a cross-quality adaptation model is developed to refine the source pose estimator using the synthesized target body parts. We perform cross-quality experiments on three datasets with different image quality using two state-of-the-art pose estimators, and compare the proposed method with five unsupervised domain adaptation methods. Our experimental results show that the proposed method outperforms not only the source pose estimators, but also other unsupervised domain adaptation methods.

Gli stili APA, Harvard, Vancouver, ISO e altri

30

Roy, Subhankar. "Learning to Adapt Neural Networks Across Visual Domains". Doctoral thesis, Università degli studi di Trento, 2022. http://hdl.handle.net/11572/354343.

Testo completo

Abstract (sommario):

In the field of machine learning (ML) a very commonly encountered problem is the lack of generalizability of learnt classification functions when subjected to new samples that are not representative of the training distribution. The discrepancy between the training (a.k.a. source) and test (a.k.a.target) distributions are caused by several latent factors such as change in appearance, illumination, viewpoints and so on, which is also popularly known as domain-shift. In order to make a classifier cope with such domain-shifts, a sub-field in machine learning called domain adaptation (DA) has emerged that jointly uses the annotated data from the source domain together with the unlabelled data from the target domain of interest. For a classifier to be adapted to an unlabelled target data set is of tremendous practical significance because it has no associated labelling cost and allows for more accurate predictions in the environment of interest. A majority of the DA methods which address the single source and single target domain scenario are not easily extendable to many practical DA scenarios. As there has been as increasing focus to make ML models deployable, it calls for devising improved methods that can handle inherently complex practical DA scenarios in the real world. In this work we build towards this goal of addressing more practical DA settings and help realize novel methods for more real world applications: (i) We begin our work with analyzing and addressing the single source and single target setting by proposing whitening-based embedded normalization layers to align the marginal feature distributions between two domains. To better utilize the unlabelled target data we propose an unsupervised regularization loss that encourages both confident and consistent predictions. (ii) Next, we build on top of the proposed normalization layers and use them in a generative framework to address multi-source DA by posing it as an image translation problem. This proposed framework TriGAN allows a single generator to be learned by using all the source domain data into a single network, leading to better generation of target-like source data. (iii) We address multi-target DA by learning a single classifier for all of the target domains. Our proposed framework exploits feature aggregation with a graph convolutional network to align feature representations of similar samples across domains. Moreover, to counteract the noisy pseudo-labels we propose to use a co-teaching strategy with a dual classifier head. To enable smoother adaptation, we propose a domain curriculum learning ,when the domain labels are available, that adapts to one target domain at a time, with increasing domain gap. (iv) Finally, we address the challenging source-free DA where the only source of supervision is a source-trained model. We propose to use Laplace Approximation to build a probabilistic source model that can quantify the uncertainty in the source model predictions on the target data. The uncertainty is then used as importance weights during the target adaptation process, down-weighting target data that do not lie in the source manifold.

Gli stili APA, Harvard, Vancouver, ISO e altri

31

Hatzichristou, Chryse, e Diether Hopf. "School adaptation of Greek children after remigration : age differences in multiple domains". Universität Potsdam, 1995. http://opus.kobv.de/ubp/volltexte/2009/1687/.

Testo completo

Abstract (sommario):

The aim of the study is to explore the patterns of adjustment of Greek remigrant children (coming from the former Federal Republic of Germany) as compared to their peers in the Greek public schools. Teacher, peer, and self-ratings were used and achievement data were obtained. The sample consisted of two age groups, elementary and secondary school students of public schools in Greece. The remigrant students were divided into two groups ("early return" and "late return") based on the year of return to Greece, and the control group consisted of all the classmates of the students. Return students were found to experience problems mainly in school performance. Contrary to the authors' hypotheses, remigrant students do not seem to experience any severe interpersonal or intrapersonal problems as compared to their local peers, indicating a rather smooth psychosocial adjustment. The authors' findings underscore the importance of the right time for remigration.

Gli stili APA, Harvard, Vancouver, ISO e altri

32

Palm, Myllylä Johannes. "Domain Adaptation for Hypernym Discovery via Automatic Collection of Domain-Speciﬁc Training Data". Thesis, Linköpings universitet, Institutionen för datavetenskap, 2019. http://urn.kb.se/resolve?urn=urn:nbn:se:liu:diva-157693.

Testo completo

Abstract (sommario):

Identifying semantic relations in natural language text is an important component of many knowledge extraction systems. This thesis studies the task of hypernym discovery, i.e discovering terms that are related by the hypernymy (is-a) relation. Speciﬁcally, this thesis explores how state-of-the-art methods for hypernym discovery perform when applied in speciﬁc language domains. In recent times, state-of-the-art methods for hypernym discovery are mostly made up by supervised machine learning models that leverage distributional word representations such as word embeddings. These models require labeled training data in the form of term pairs that are known to be related by hypernymy. Such labeled training data is often not available when working with a speciﬁc language domain. This thesis presents experiments with an automatic training data collection algorithm. The algorithm leverages a pre-deﬁned domain-speciﬁc vocabulary, and the lexical resource WordNet, to extract training pairs automatically. This thesis contributes by presenting experimental results when attempting to leverage such automatically collected domain-speciﬁc training data for the purpose of domain adaptation. Experiments are conducted in two different domains: One domain where there is a large amount of text data, and another domain where there is a much smaller amount of text data. Results show that the automatically collected training data has a positive impact on performance in both domains. The performance boost is most signiﬁcant in the domain with a large amount of text data, with mean average precision increasing by up to 8 points.

Gli stili APA, Harvard, Vancouver, ISO e altri

33

Donati, Lorenzo. "Domain Adaptation through Deep Neural Networks for Health Informatics". Master's thesis, Alma Mater Studiorum - Università di Bologna, 2017. http://amslaurea.unibo.it/14888/.

Testo completo

Abstract (sommario):

The PreventIT project is an EU Horizon 2020 project aimed at preventing early functional decline at younger old age. The analysis of causal links between risk factors and functional decline has been made possible by the cooperation of several research institutes' studies. However, since each research institute collects and delivers different kinds of data in different formats, so far the analysis has been assisted by expert geriatricians whose role is to detect the best candidates among hundreds of fields and offer a semantic interpretation of the values. This manual data harmonization approach is very common in both scientific and industrial environments. In this thesis project an alternative method for parsing heterogeneous data is proposed. Since all the datasets represent semantically related data, being all made from longitudinal studies on aging-related metrics, it is possible to train an artificial neural network to perform an automatic domain adaptation. To achieve this goal, a Stacked Denoising Autoencoder has been implemented and trained to extract a domain-invariant representation of the data. Then, from this high-level representation, multiple classifiers have been trained to validate the model and ultimately to predict the probability of functional decline of the patient. This innovative approach to the domain adaptation process can provide an easy and fast solution to many research fields that now rely on human interaction to analyze the semantic data model and perform cross-dataset analysis. Functional decline classifiers show a great improvement in their performance when trained on the domain-invariant features extracted by the Stacked Denoising Autoencoder. Furthermore, this project applies multiple deep neural network classifiers on top of the Stacked Denoising Autoencoder representation, achieving excellent results for the prediction of functional decline in a real case study that involves two different datasets.

Gli stili APA, Harvard, Vancouver, ISO e altri

34

Margineanu, Elena. "Institutional adaptation in environmental domain: the case of Moldova". Thesis, Тернопіль: Вектор, 2020. http://er.nau.edu.ua/handle/NAU/41786.

Testo completo

Abstract (sommario):

Environment currently, more than ever, receives high attention as (for example) scandalous cases of disturbed ecosystem due to negligent waste management system are widely distributed through media and social channels. Even if we are making abstraction of social activists that protest against, as they claim, climate change, it is true that most governments acknowledge the pressure to introduce new measures and tools for tackling environmental issues with its broad spectrum of sub-domains.

Gli stili APA, Harvard, Vancouver, ISO e altri

35

Manamasa, Krishna Himaja. "Domain adaptation from 3D synthetic images to real images". Thesis, Blekinge Tekniska Högskola, Institutionen för datavetenskap, 2020. http://urn.kb.se/resolve?urn=urn:nbn:se:bth-19303.

Testo completo

Abstract (sommario):

Background. Domain adaptation is described as, a model learning from a source data distribution and performing well on the target data. This concept, Domain adaptation is applied to assembly-line production tasks to perform an automatic quality inspection. Objectives. The aim of this master thesis is to apply this concept of 3D domain adaptation from synthetic images to real images. It is an attempt to bridge the gap between different domains (synthetic and real point cloud images), by implementing deep learning models that learn from synthetic 3D point cloud (CAD model images) and perform well on the actual 3D point cloud (3D Camera images). Methods. Through this course of thesis project, various methods for understand- ing the data and analyzing it for bridging the gap between CAD and CAM to make them similar is looked into. Literature review and controlled experiment are research methodologies followed during implementation. In this project, we experiment with four different deep learning models with data generated and compare their performance to know which deep learning model performs best for the data. Results. The results are explained through metrics i.e, accuracy and train time, which were the outcomes of each of the deep learning models after the experiment. These metrics are illustrated in the form of graphs for comparative analysis between the models on which the data is trained and tested on. PointDAN showed better results with higher accuracy compared to the other 3 models. Conclusions. The results attained show that domain adaptation for synthetic images to real images is possible with the data generated. PointDAN deep learning model which focuses on local feature alignment and global feature alignment with single-view point data shows better results with our data.

Gli stili APA, Harvard, Vancouver, ISO e altri

36

Liu, Ye. "Application of Convolutional Deep Belief Networks to Domain Adaptation". The Ohio State University, 2014. http://rave.ohiolink.edu/etdc/view?acc_num=osu1397728737.

Testo completo

Gli stili APA, Harvard, Vancouver, ISO e altri

37

Chinea, Ríos Mara. "Advanced techniques for domain adaptation in Statistical Machine Translation". Doctoral thesis, Universitat Politècnica de València, 2019. http://hdl.handle.net/10251/117611.

Testo completo

Abstract (sommario):

[ES] La Traducción Automática Estadística es un sup-campo de la lingüística computacional que investiga como emplear los ordenadores en el proceso de traducción de un texto de un lenguaje humano a otro. La traducción automática estadística es el enfoque más popular que se emplea para construir estos sistemas de traducción automáticos. La calidad de dichos sistemas depende en gran medida de los ejemplos de traducción que se emplean durante los procesos de entrenamiento y adaptación de los modelos. Los conjuntos de datos empleados son obtenidos a partir de una gran variedad de fuentes y en muchos casos puede que no tengamos a mano los datos más adecuados para un dominio específico. Dado este problema de carencia de datos, la idea principal para solucionarlo es encontrar aquellos conjuntos de datos más adecuados para entrenar o adaptar un sistema de traducción. En este sentido, esta tesis propone un conjunto de técnicas de selección de datos que identifican los datos bilingües más relevantes para una tarea extraídos de un gran conjunto de datos. Como primer paso en esta tesis, las técnicas de selección de datos son aplicadas para mejorar la calidad de la traducción de los sistemas de traducción bajo el paradigma basado en frases. Estas técnicas se basan en el concepto de representación continua de las palabras o las oraciones en un espacio vectorial. Los resultados experimentales demuestran que las técnicas utilizadas son efectivas para diferentes lenguajes y dominios. El paradigma de Traducción Automática Neuronal también fue aplicado en esta tesis. Dentro de este paradigma, investigamos la aplicación que pueden tener las técnicas de selección de datos anteriormente validadas en el paradigma basado en frases. El trabajo realizado se centró en la utilización de dos tareas diferentes de adaptación del sistema. Por un lado, investigamos cómo aumentar la calidad de traducción del sistema, aumentando el tamaño del conjunto de entrenamiento. Por otro lado, el método de selección de datos se empleó para crear un conjunto de datos sintéticos. Los experimentos se realizaron para diferentes dominios y los resultados de traducción obtenidos son convincentes para ambas tareas. Finalmente, cabe señalar que las técnicas desarrolladas y presentadas a lo largo de esta tesis pueden implementarse fácilmente dentro de un escenario de traducción real.
[CAT] La Traducció Automàtica Estadística és un sup-camp de la lingüística computacional que investiga com emprar els ordinadors en el procés de traducció d'un text d'un llenguatge humà a un altre. La traducció automàtica estadística és l'enfocament més popular que s'empra per a construir aquests sistemes de traducció automàtics. La qualitat d'aquests sistemes depèn en gran mesura dels exemples de traducció que s'empren durant els processos d'entrenament i adaptació dels models. Els conjunts de dades emprades són obtinguts a partir d'una gran varietat de fonts i en molts casos pot ser que no tinguem a mà les dades més adequades per a un domini específic. Donat aquest problema de manca de dades, la idea principal per a solucionar-ho és trobar aquells conjunts de dades més adequades per a entrenar o adaptar un sistema de traducció. En aquest sentit, aquesta tesi proposa un conjunt de tècniques de selecció de dades que identifiquen les dades bilingües més rellevants per a una tasca extrets d'un gran conjunt de dades. Com a primer pas en aquesta tesi, les tècniques de selecció de dades són aplicades per a millorar la qualitat de la traducció dels sistemes de traducció sota el paradigma basat en frases. Aquestes tècniques es basen en el concepte de representació contínua de les paraules o les oracions en un espai vectorial. Els resultats experimentals demostren que les tècniques utilitzades són efectives per a diferents llenguatges i dominis. El paradigma de Traducció Automàtica Neuronal també va ser aplicat en aquesta tesi. Dins d'aquest paradigma, investiguem l'aplicació que poden tenir les tècniques de selecció de dades anteriorment validades en el paradigma basat en frases. El treball realitzat es va centrar en la utilització de dues tasques diferents. D'una banda, investiguem com augmentar la qualitat de traducció del sistema, augmentant la grandària del conjunt d'entrenament. D'altra banda, el mètode de selecció de dades es va emprar per a crear un conjunt de dades sintètiques. Els experiments es van realitzar per a diferents dominis i els resultats de traducció obtinguts són convincents per a ambdues tasques. Finalment, cal assenyalar que les tècniques desenvolupades i presentades al llarg d'aquesta tesi poden implementar-se fàcilment dins d'un escenari de traducció real.
[EN] La Traducció Automàtica Estadística és un sup-camp de la lingüística computacional que investiga com emprar els ordinadors en el procés de traducció d'un text d'un llenguatge humà a un altre. La traducció automàtica estadística és l'enfocament més popular que s'empra per a construir aquests sistemes de traducció automàtics. La qualitat d'aquests sistemes depèn en gran mesura dels exemples de traducció que s'empren durant els processos d'entrenament i adaptació dels models. Els conjunts de dades emprades són obtinguts a partir d'una gran varietat de fonts i en molts casos pot ser que no tinguem a mà les dades més adequades per a un domini específic. Donat aquest problema de manca de dades, la idea principal per a solucionar-ho és trobar aquells conjunts de dades més adequades per a entrenar o adaptar un sistema de traducció. En aquest sentit, aquesta tesi proposa un conjunt de tècniques de selecció de dades que identifiquen les dades bilingües més rellevants per a una tasca extrets d'un gran conjunt de dades. Com a primer pas en aquesta tesi, les tècniques de selecció de dades són aplicades per a millorar la qualitat de la traducció dels sistemes de traducció sota el paradigma basat en frases. Aquestes tècniques es basen en el concepte de representació contínua de les paraules o les oracions en un espai vectorial. Els resultats experimentals demostren que les tècniques utilitzades són efectives per a diferents llenguatges i dominis. El paradigma de Traducció Automàtica Neuronal també va ser aplicat en aquesta tesi. Dins d'aquest paradigma, investiguem l'aplicació que poden tenir les tècniques de selecció de dades anteriorment validades en el paradigma basat en frases. El treball realitzat es va centrar en la utilització de dues tasques diferents d'adaptació del sistema. D'una banda, investiguem com augmentar la qualitat de traducció del sistema, augmentant la grandària del conjunt d'entrenament. D'altra banda, el mètode de selecció de dades es va emprar per a crear un conjunt de dades sintètiques. Els experiments es van realitzar per a diferents dominis i els resultats de traducció obtinguts són convincents per a ambdues tasques. Finalment, cal assenyalar que les tècniques desenvolupades i presentades al llarg d'aquesta tesi poden implementar-se fàcilment dins d'un escenari de traducció real.
Chinea Ríos, M. (2019). Advanced techniques for domain adaptation in Statistical Machine Translation [Tesis doctoral no publicada]. Universitat Politècnica de València. https://doi.org/10.4995/Thesis/10251/117611
TESIS

Gli stili APA, Harvard, Vancouver, ISO e altri

38

Radhakrishnan, Saieshwar. "Domain Adaptation of IMU sensors using Generative Adversarial Networks". Thesis, KTH, Skolan för elektroteknik och datavetenskap (EECS), 2020. http://urn.kb.se/resolve?urn=urn:nbn:se:kth:diva-286821.

Testo completo

Abstract (sommario):

Autonomous vehicles rely on sensors for a clear understanding of the environment and in a heavy duty truck, the sensors are placed at multiple locations like the cabin, chassis and the trailer in order to increase the field of view and reduce the blind spot area. Usually, these sensors perform best when they are stationary relative to the ground, hence large and fast movements, which are quite common in a truck, may lead to performance reduction, erroneous data or in the worst case, a sensor failure. This enforces a need to validate the sensors before using them for making life-critical decisions. This thesis proposes Domain Adaptation as one of the strategies to co-validate Inertial Measurement Unit (IMU) sensors. The proposed Generative Adversarial Network (GAN) based framework predicts the data of one IMU using other IMUs in the truck by implicitly learning the internal dynamics. This prediction model along with other sensor fusion strategies would be used by the supervising system to validate the IMUs in real-time. Through data collected from real-world experiments, it is shown that the proposed framework is able to accurately transform raw IMU sequences across domains. A further comparison is made between Long Short Term Memory (LSTM) and WaveNet based architectures to show the superiority of WaveNets in terms of performance and computational efficiency.
Autonoma fordon förlitar sig på sensorer för att skapa en bild av omgivningen. På en tung lastbil placeras sensorerna på multipla ställen, till exempel på hytten, chassiet och på trailern för att öka siktfältet och för att minska blinda områden. Vanligtvis presterar sensorerna som bäst när de är stationära i förhållande till marken, därför kan stora och snabba rörelser, som är vanliga på en lastbil, leda till nedsatt prestanda, felaktig data och i värsta fall fallerande sensorer. På grund av detta så finns det ett stort behov av att validera sensordata innan det används för kritiskt beslutsfattande. Den här avhandlingen föreslår domänadaption som en av de strategier för att samvalidera Tröghetsmätningssensorer (IMU-sensorer). Det föreslagna Generative Adversarial Network (GAN) baserade ramverket förutspår en Tröghetssensors data genom att implicit lära sig den interna dynamiken från andra Tröghetssensorer som är monterade på lastbilen. Den här prediktionsmodellen kombinerat med andra sensorfusionsstrategier kan användas av kontrollsystemet för att i realtid validera Tröghetssensorerna. Med hjälp av data insamlat från verkliga experiment visas det att det föreslagna ramverket klarar av att med hög noggrannhet konvertera obehandlade Tröghetssensor-sekvenser mellan domäner. Ytterligare en undersökning mellan Long Short Term Memory (LSTM) och WaveNet-baserade arkitekturer görs för att visa överlägsenheten i WaveNets när det gäller prestanda och beräkningseffektivitet.

Gli stili APA, Harvard, Vancouver, ISO e altri

39

Zhang, Xinwen. "Multi-modality Medical Image Segmentation with Unsupervised Domain Adaptation". Thesis, The University of Sydney, 2022. https://hdl.handle.net/2123/29776.

Testo completo

Abstract (sommario):

Advances in medical imaging have greatly aided in providing accurate and fast medical diagnosis, followed by recent deep learning developments enabling the efficient and cost-effective analysis of medical images. Among different image processing tasks, medical segmentation is one of the most crucial aspects because it provides the class, location, size, and shape of the subject of interest, which is invaluable and essential for diagnostics. Nevertheless, acquiring annotations for training data usually requires expensive manpower and specialised expertise, making supervised training difficult. To overcome these problems, unsupervised domain adaptation (UDA) has been adopted to bridge knowledge between different domains. Despite the appearance dissimilarities of different modalities such as MRI and CT, researchers have concluded that structural features of the same anatomy are universal across modalities, which unfolded the study of multi-modality image segmentation with UDA methods. The traditional UDA research tackled the domain shift problem by minimising the distance of the source and target distributions in latent spaces with the help of advanced mathematics. However, with the recent development of the generative adversarial network (GAN), the adversarial UDA methods have shown outstanding performance by producing synthetic images to mitigate the domain gap in training a segmentation network for the target domain. Most existing studies focus on modifying the network architecture, but few investigate the generative adversarial training strategy. Inspired by the recent success of state-of-the-art data augmentation techniques in classification tasks, we designed a novel mix-up strategy to assist GAN training for the better synthesis of structural details, consequently leading to better segmentation results. In this thesis, we propose SynthMix, an add-on module with a natural yet effective training policy that can promote synthetic quality without altering the network architecture. SynthMix is a mix-up synthesis scheme designed for integration with the adversarial logic of GAN networks. Traditional GAN approaches judge an image as a whole which could be easily dominated by discriminative features, resulting in little improvement of delicate structures in synthesis. In contrast, SynthMix uses the data augmentation technique to reinforce detail transformation at local regions. Specifically, it coherently mixes up aligned images of real and synthetic samples at local regions to stimulate the generation of fine-grained features examined by an associated inspector for domain-specific details. We evaluated our method on two segmentation benchmarks among three publicly available datasets. Our method showed a significant performance gain compared with existing state-of-the-art approaches.

Gli stili APA, Harvard, Vancouver, ISO e altri

40

Tian, Tian. "Domain Adaptation and Model Combination for the Annotation of Multi-source, Multi-domain Texts". Thesis, Paris 3, 2019. http://www.theses.fr/2019PA030003.

Testo completo

Abstract (sommario):

Internet propose aujourd’hui aux utilisateurs de services en ligne de commenter, d’éditer et de partager leurs points de vue sur différents sujets de discussion. Ce type de contenu est maintenant devenu la ressource principale pour les analyses d’opinions sur Internet. Néanmoins, à cause des abréviations, du bruit, des fautes d’orthographe et toutes autres sortes de problèmes, les outils de traitements automatiques des langues, y compris les reconnaisseurs d’entités nommées et les étiqueteurs automatiques morphosyntaxiques, ont des performances plus faibles que sur les textes bien-formés (Ritter et al., 2011).Cette thèse a pour objet la reconnaissance d’entités nommées sur les contenus générés par les utilisateurs sur Internet. Nous avons établi un corpus d’évaluation avec des textes multi-sources et multi-domaines. Ensuite, nous avons développé un modèle de champs conditionnels aléatoires, entrainé sur un corpus annoté provenant des contenus générés par les utilisateurs.Dans le but d’améliorer les résultats de la reconnaissance d’entités nommées, nous avons d’abord développé un étiqueteur morpho-syntaxique sur les contenus générés par les utilisateurs et nous avons utilisé les étiquettesprédites comme un attribut du modèle des champs conditionnels aléatoire. Enfin, pour transformer les contenus générés par les utilisateurs en textes bien-formés, nous avons développé un modèle de normalisation lexicale basé sur des réseaux de neurones pour proposer une forme correcte pour les mots non-standard
The increasing mass of User-Generated Content (UGC) on the Internet means that people are now willing to comment, edit or share their opinions on different topics. This content is now the main ressource for sentiment analysis on the Internet. Due to abbreviations, noise, spelling errors and all other problems with UGC, traditional Natural Language Processing (NLP) tools, including Named Entity Recognizers and part-of-speech (POS) taggers, perform poorly when compared to their usual results on canonical text (Ritter et al., 2011).This thesis deals with Named Entity Recognition (NER) on some User-Generated Content (UGC). We have created an evaluation dataset including multi-domain and multi-sources texts. We then developed a Conditional Random Fields (CRFs) model trained on User-Generated Content (UGC).In order to improve NER results in this context, we first developed a POStagger on UGC and used the predicted POS tags as a feature in the CRFs model. To turn UGC into canonical text, we also developed a normalization model using neural networks to propose a correct form for Non-Standard Words (NSW) in the UGC
各种社交网络应用使得互联网用户对各种话题的实时评价，编辑和分享成为可能。这类用户生成的文本内容(User Generated content)已成为社交网络上意见分析的主要目标和来源。但是，此类文本内容中包含的缩写，噪声（不规则词），拼写错误以及其他各种问题导致包括命名实体识别，词性标注在内的传统的自然语言处理工具的性能，相比良好组成的文本降低了许多【参见Ritter 2011】。本论文的主要目标是针对社交网络上用户生成文本内容的命名实体识别。我们首先建立了一个包含多来源，多领域文本的有标注的语料库作为标准评价语料库。然后，我们开发了一个由社交网络用户生成文本训练的基于条件随机场(Conditional Random Fields)的序列标注模型。基于改善这个命名实体识别模型的目的，我们又开发了另一个同样由社交网络用户生成内容训练的词性标注模型，并使用此模型预测的词性作为命名实体识别的条件随机场模型的特征。最后，为了将用户生成文本内容转换成相对标准的良好文本内容，我们开发了一个基于神经网络的词汇标准化模型，用以改正用户生成文本内容中的不标准字，并使用模型提供的改正形式作为命名实体识别的条件随机场模型的特征，借以改善原模型的性能。

Gli stili APA, Harvard, Vancouver, ISO e altri

41

Vázquez, Bermúdez David. "Domain Adaptation of Virtual and Real Worlds for Pedestrian Detection". Doctoral thesis, Universitat Autònoma de Barcelona, 2013. http://hdl.handle.net/10803/125977.

Testo completo

Abstract (sommario):

La detección de peatones es clave para muchas aplicaciones como asistencia al conductor, video vigilancia o multimedia. Los mejores detectores se basan en clasificadores basados en modelos de apariencia entrenados con ejemplos anotados. Sin embargo, el proceso de anotación es una tarea intensiva y subjetiva cuando es llevada a cabo por personas. Por ello, vale la pena minimizar la intervención humana en dicha tarea mediante el uso de herramientas computacionales como los mundos virtuales porque con ellos podemos obtener anotaciones variadas y precisas de forma rápida. Sin embargo, el uso de este tipo de datos genera la siguiente pregunta: ¿Es posible que un modelo de apariencia entrenado en un mundo virtual pueda funcionar de manera satisfactoria en el mundo real? Para responder esta pregunta, hemos realizado diferentes experimentos que sugieren que los clasificadores entrenados en el mundo virtual pueden ofrecer buenos resultados al aplicarse en ambientes del mundo real. Sin embargo, también se encontró que en algunos casos estos clasificadores se pueden ver afectados por el problema conocido como el cambio en la naturaleza de los datos, igual que ocurre con los clasificadores entrenados en el mundo real. En consecuencia, hemos diseñado un sistema de adaptación de dominio, V-AYLA, en el que hemos probado diferentes técnicas para recoger unos pocos ejemplos del mundo real y combinarlos con una gran cantidad de ejemplos del mundo virtual para entrenar un detector de peatones adaptado. V-AYLA ofrece la misma precisión de detección que un detector entrenado con anotaciones manuales y probado con imágenes reales del mismo dominio. Idealmente, nos gustaría que nuestro sistema se adaptase automáticamente sin necesidad de intervenci ón humana. Por ello, a modo de demostración, proponemos utilizar técnicas de adaptación no supervisadas que permitan eliminar completamente la intervención humana del proceso de adaptación. Hasta donde sabemos, este es el primer trabajo que muestra que es posible desarrollar un detector de objetos en el mundo virtual y adaptarlo al mundo real. Finalmente, proponemos una estrategia diferente para evitar el problema del cambio en la naturaleza de los datos que consiste en recoger ejemplos en el mundo real y reentrenar solamente con ellos pero haciéndolo de tal modo que no se tengan que anotar peatones en el mundo real. El resultado de este clasificador es equivalente a otro entrenado con anotaciones obtenidas de forma manual. Los resultados presentados en esta tesis no se limitan a adaptar un detector de peatones virtuales al mundo real, sino que va más allá, mostrando una nueva metodología que permitiría a un sistema adaptarse a cualquier nueva situación y que sienta las bases para la investigación futura en este campo todavía sin explorar.
Pedestrian detection is of paramount interest for many applications, e.g. Advanced Driver Assistance Systems, Surveillance and Media. Most promising pedestrian detectors rely on appearance-based classifiers trained with annotated samples. However, the required annotation step represents an intensive and subjective task when it has to be done by persons. Therefore, it is worth to minimize the human intervention in such a task by using computational tools like realistic virtual worlds, where precise and rich annotations of visual information can be automatically generated. Nevertheless, the use of this kind of data generates the following question: can a pedestrian appearance model learnt with virtual-world data work successfully for pedestrian detection in real- world scenarios?. To answer this question, we conducted different experiments that suggest that classifiers based on virtual-world data can perform well in real-world environments. However, it was also found that in some cases these classifiers can suffer the so called dataset shift problem as real-world based classifiers does. Accordingly, we have designed a domain adaptation framework, V-AYLA, in which we have explored different techniques to collect a few pedestrian samples from the target domain (real world) and combine them with many samples of the source domain (virtual world) in order to train a domain adapted pedestrian classifier. V-AYLA reports the same detection performance as the one obtained by training with human-provided pedestrian annotations and testing with real-world images from the same domain. Ideally, we would like to adapt our system without any human intervention. Therefore, as a first proof of concept we proposed the use of an unsupervised domain adaptation technique that avoids human intervention during the adaptation process. To the best of our knowledge, this is the first work that demonstrates adaptation of virtual and real worlds for developing an object detector. We also assess a different strategy to avoid the dataset shift that consists in collecting real-world samples and retrain with them, but in such a way that no bounding boxes of real-world pedestrians have to be provided. We show that the generated classifier is competitive with respect to the counterpart trained with samples collected by manually annotating pedestrian bounding boxes. The results presented on this Thesis not only end with a proposal for adapting a virtual-world pedestrian detector to the real world, but also it goes further by pointing out a new methodology that would allow the system to adapt to different situations, which we hope will provide the foundations for future research in this unexplored area.

Gli stili APA, Harvard, Vancouver, ISO e altri

42

Ruberg, Anders. "Frequency Domain Link Adaptation for OFDM-based Cellular Packet Data". Thesis, Linköping University, Department of Electrical Engineering, 2006. http://urn.kb.se/resolve?urn=urn:nbn:se:liu:diva-6328.

Testo completo

Abstract (sommario):

In order to be competitive with emerging mobile systems and to satisfy the ever growing request for higher data rates, the 3G consortium, 3rd Generation Partnership Project (3GPP), is currently developing concepts for a long term evolution (LTE) of the 3G standard. The LTE-concept at Ericsson is based on Orthogonal Frequency Division Multiplexing (OFDM) as downlink air interface. OFDM enables the use of frequency domain link adaptation to select the most appropriate transmission parameters according to current channel conditions, in order to maximize the throughput and maintain the delay at a desired level. The purpose of this thesis work is to study, implement and evaluate different link adaptation algorithms. The main focus is on modulation adaptation, where the differences in performance between time domain and frequency domain adaptation are investigated. The simulations made in this thesis are made with a simulator developed at Ericsson. Simulations show in general that the cell throughput is enhanced by an average of 3% when using frequency domain modulation adaptation. When using the implemented frequency domain power allocation algorithm, a gain of 23-36% in average is seen in the users 5th percentile throughput. It should be noted that the simulations use a realistic web traffic model, which makes the channel quality estimation (CQE) difficult. The CQE has great impact on the performance of frequency domain adaptation. Throughput improvements are expected when using an improved CQE or interference avoidance schemes. The gains with frequency domain adaptation shown in this thesis work may be too small to motivate the extra signalling overhead required. The complexity of the implemented frequency domain power allocation algorithm is also very high compared to the performance enhancement seen.

Gli stili APA, Harvard, Vancouver, ISO e altri

43

XIAO, MIN. "Generalized Domain Adaptation for Sequence Labeling in Natural Language Processing". Diss., Temple University Libraries, 2016. http://cdm16002.contentdm.oclc.org/cdm/ref/collection/p245801coll10/id/391382.

Testo completo

Abstract (sommario):

Computer and Information Science
Ph.D.
Sequence labeling tasks have been widely studied in the natural language processing area, such as part-of-speech tagging, syntactic chunking, dependency parsing, and etc. Most of those systems are developed on a large amount of labeled training data via supervised learning. However, manually collecting labeled training data is too time-consuming and expensive. As an alternative, to alleviate the issue of label scarcity, domain adaptation has recently been proposed to train a statistical machine learning model in a target domain where there is no enough labeled training data by exploiting existing free labeled training data in a different but related source domain. The natural language processing community has witnessed the success of domain adaptation in a variety of sequence labeling tasks. Though the labeled training data in the source domain are available and free, however, they are not exactly as and can be very different from the test data in the target domain. Thus, simply applying naive supervised machine learning algorithms without considering domain differences may not fulfill the purpose. In this dissertation, we developed several novel representation learning approaches to address domain adaptation for sequence labeling in natural language processing. Those representation learning techniques aim to induce latent generalizable features to bridge domain divergence to enable cross-domain prediction. We first tackle a semi-supervised domain adaptation scenario where the target domain has a small amount of labeled training data and propose a distributed representation learning approach based on a probabilistic neural language model. We then relax the assumption of the availability of labeled training data in the target domain and study an unsupervised domain adaptation scenario where the target domain has only unlabeled training data, and give a task-informative representation learning approach based on dynamic dependency networks. Both works are developed in the setting where different domains contain sentences in different genres. We then extend and generalize domain adaptation into a more challenging scenario where different domains contain sentences in different languages and propose two cross-lingual representation learning approaches, one is based on deep neural networks with auxiliary bilingual word pairs and the other is based on annotation projection with auxiliary parallel sentences. All four specific learning scenarios are extensively evaluated with different sequence labeling tasks. The empirical results demonstrate the effectiveness of those generalized domain adaptation techniques for sequence labeling in natural language processing.
Temple University--Theses

Gli stili APA, Harvard, Vancouver, ISO e altri

44

Xu, Brian(Brian W. ). "Combating fake news with adversarial domain adaptation and neural models". Thesis, Massachusetts Institute of Technology, 2019. https://hdl.handle.net/1721.1/121689.

Testo completo

Abstract (sommario):

This electronic version was submitted by the student author. The certified thesis is available in the Institute Archives and Special Collections.
Thesis: M. Eng., Massachusetts Institute of Technology, Department of Electrical Engineering and Computer Science, 2019
Cataloged from student-submitted PDF version of thesis.
Includes bibliographical references (pages 77-80).
Factually incorrect claims on the web and in social media can cause considerable damage to individuals and societies by misleading them. As we enter an era where it is easier than ever to disseminate "fake news" and other dubious claims, automatic fact checking becomes an essential tool to help people discern fact from fiction. In this thesis, we focus on two main tasks: fact checking which involves classifying an input claim with respect to its veracity, and stance detection which involves determining the perspective of a document with respect to a claim. For the fact checking task, we present Bidirectional Long Short Term Memory (Bi-LSTM) and Convolutional Neural Network (CNN) based models and conduct our experiments on the LIAR dataset [Wang, 2017], a recently released fact checking task. Our model outperforms the state of the art baseline on this dataset. For the stance detection task, we present bag of words (BOW) and CNN based models in hierarchy schemes. These architectures are then supplemented with an adversarial domain adaptation technique, which helps the models overcome dataset size limitations. We test the performance of these models by using the Fake News Challenge (FNC) [Pomerleau and Rao, 2017], the Fact Extraction and VERification (FEVER) [Thorne et al., 2018], and the Stanford Natural Language Inference (SNLI) [Bowman et al., 2015] datasets. Our experiments yielded a model which has state of the art performance on FNC target data by using FEVER source data coupled with adversarial domain adaptation [Xu et al., 2018].
by Brian Xu.
M. Eng.
M.Eng. Massachusetts Institute of Technology, Department of Electrical Engineering and Computer Science

Gli stili APA, Harvard, Vancouver, ISO e altri

45

Peyrache, Jean-Philippe. "Nouvelles approches itératives avec garanties théoriques pour l'adaptation de domaine non supervisée". Thesis, Saint-Etienne, 2014. http://www.theses.fr/2014STET4023/document.

Testo completo

Abstract (sommario):

Ces dernières années, l’intérêt pour l’apprentissage automatique n’a cessé d’augmenter dans des domaines aussi variés que la reconnaissance d’images ou l’analyse de données médicales. Cependant, une limitation du cadre classique PAC a récemment été mise en avant. Elle a entraîné l’émergence d’un nouvel axe de recherche : l’Adaptation de Domaine, dans lequel on considère que les données d’apprentissage proviennent d’une distribution (dite source) différente de celle (dite cible) dont sont issues les données de test. Les premiers travaux théoriques effectués ont débouché sur la conclusion selon laquelle une bonne performance sur le test peut s’obtenir en minimisant à la fois l’erreur sur le domaine source et un terme de divergence entre les deux distributions. Trois grandes catégories d’approches s’en inspirent : par repondération, par reprojection et par auto-étiquetage. Dans ce travail de thèse, nous proposons deux contributions. La première est une approche de reprojection basée sur la théorie du boosting et s’appliquant aux données numériques. Celle-ci offre des garanties théoriques intéressantes et semble également en mesure d’obtenir de bonnes performances en généralisation. Notre seconde contribution consiste d’une part en la proposition d’un cadre permettant de combler le manque de résultats théoriques pour les méthodes d’auto-étiquetage en donnant des conditions nécessaires à la réussite de ce type d’algorithme. D’autre part, nous proposons dans ce cadre une nouvelle approche utilisant la théorie des (epsilon, gamma, tau)-bonnes fonctions de similarité afin de contourner les limitations imposées par la théorie des noyaux dans le contexte des données structurées
During the past few years, an increasing interest for Machine Learning has been encountered, in various domains like image recognition or medical data analysis. However, a limitation of the classical PAC framework has recently been highlighted. It led to the emergence of a new research axis: Domain Adaptation (DA), in which learning data are considered as coming from a distribution (the source one) different from the one (the target one) from which are generated test data. The first theoretical works concluded that a good performance on the target domain can be obtained by minimizing in the same time the source error and a divergence term between the two distributions. Three main categories of approaches are derived from this idea : by reweighting, by reprojection and by self-labeling. In this thesis work, we propose two contributions. The first one is a reprojection approach based on boosting theory and designed for numerical data. It offers interesting theoretical guarantees and also seems able to obtain good generalization performances. Our second contribution consists first in a framework filling the gap of the lack of theoretical results for self-labeling methods by introducing necessary conditions ensuring the good behavior of this kind of algorithm. On the other hand, we propose in this framework a new approach, using the theory of (epsilon, gamma, tau)- good similarity functions to go around the limitations due to the use of kernel theory in the specific context of structured data

Gli stili APA, Harvard, Vancouver, ISO e altri

46

Panareda, Busto Pau [Verfasser]. "Domain Adaptation for Image Recognition and Viewpoint Estimation / Pau Panareda Busto". Bonn : Universitäts- und Landesbibliothek Bonn, 2020. http://d-nb.info/1219140449/34.

Testo completo

Gli stili APA, Harvard, Vancouver, ISO e altri

47

Nyströmer, Carl. "Musical Instrument Activity Detection using Self-Supervised Learning and Domain Adaptation". Thesis, KTH, Skolan för elektroteknik och datavetenskap (EECS), 2020. http://urn.kb.se/resolve?urn=urn:nbn:se:kth:diva-280810.

Testo completo

Abstract (sommario):

With the ever growing media and music catalogs, tools that search and navigate this data are important. For more complex search queries, meta-data is needed, but to manually label the vast amounts of new content is impossible. In this thesis, automatic labeling of musical instrument activities in song mixes is investigated, with a focus on ways to alleviate the lack of annotated data for instrument activity detection models. Two methods for alleviating the problem of small amounts of data are proposed and evaluated. Firstly, a self-supervised approach based on automatic labeling and mixing of randomized instrument stems is investigated. Secondly, a domain-adaptation approach that trains models on sampled MIDI files for instrument activity detection on recorded music is explored. The self-supervised approach yields better results compared to the baseline and points to the fact that deep learning models can learn instrument activity detection without an intrinsic musical structure in the audio mix. The domain-adaptation models trained solely on sampled MIDI files performed worse than the baseline, however using MIDI data in conjunction with recorded music boosted the performance. A hybrid model combining both self-supervised learning and domain adaptation by using both sampled MIDI data and recorded music produced the best results overall.
I och med de ständigt växande media- och musikkatalogerna krävs verktyg för att söka och navigera i dessa. För mer komplexa sökförfrågningar så behövs det metadata, men att manuellt annotera de enorma mängderna av ny data är omöjligt. I denna uppsats undersöks automatisk annotering utav instrumentsaktivitet inom musik, med ett fokus på bristen av annoterad data för modellerna för instrumentaktivitetsigenkänning. Två metoder för att komma runt bristen på data föreslås och undersöks. Den första metoden bygger på självövervakad inlärning baserad på automatisk annotering och slumpartad mixning av olika instrumentspår. Den andra metoden använder domänadaption genom att träna modeller på samplade MIDI-filer för detektering av instrument i inspelad musik. Metoden med självövervakning gav bättre resultat än baseline och pekar på att djupinlärningsmodeller kan lära sig instrumentigenkänning trots att ljudmixarna saknar musikalisk struktur. Domänadaptionsmodellerna som endast var tränade på samplad MIDI-data presterade sämre än baseline, men att använda MIDI-data tillsammans med data från inspelad musik gav förbättrade resultat. En hybridmodell som kombinerade både självövervakad inlärning och domänadaption genom att använda både samplad MIDI-data och inspelad musik gav de bästa resultaten totalt.

Gli stili APA, Harvard, Vancouver, ISO e altri

48

Sedinkina, Marina [Verfasser], e Hinrich [Akademischer Betreuer] Schütze. "Domain adaptation in Natural Language Processing / Marina Sedinkina ; Betreuer: Hinrich Schütze". München : Universitätsbibliothek der Ludwig-Maximilians-Universität, 2021. http://d-nb.info/1233966936/34.

Testo completo

Gli stili APA, Harvard, Vancouver, ISO e altri

49

Zen, Gloria. "Understanding Visual Information: from Unsupervised Discovery to Minimal Effort Domain Adaptation". Doctoral thesis, Università degli studi di Trento, 2015. https://hdl.handle.net/11572/368625.

Testo completo

Abstract (sommario):

Visual data interpretation is a fascinating problem which has received an increasing attention in the last decades. Reasons for this growing trend can be found within multiple interconnected factors, such as the exponential growth of visual data (e.g. images and videos) availability, the consequent demand for an automatic way to interpret these data and the increase of computational power. In a supervised machine learning approach, a large effort within the research community has been devoted to the collection of training samples to be provided to the learning system, resulting in the generation of very large scale datasets. This has lead to remarkable performance advances in tasks such as scene recognition or object detection, however, at a considerable high cost in terms of human labeling effort. In light of the labeling cost issue, together with the dataset bias one, another significant research direction was headed towards developing methods for learning without or with a limited amount of training data, by leveraging instead on data properties like intrinsic redundancy, time constancy or commonalities shared among different domains. Our work is in line with this last type of approach. In particular, by covering different case scenarios - from dynamic crowded scenes to facial expression analysis - we propose a novel approach to overcome some of the state-of-the-art limitations. Based on the renowned bag of words (BoW) approach, we propose a novel method which achieves higher performances in tasks such as learning typical patterns of behaviors and anomalies discovery from complex scenes, by considering the similarity among visual words in the learning phase. We also show that including sparsity constraints can help dealing with noise which is intrinsic to low level cues extracted from complex dynamic scenes. Facing the so called dataset bias issue, we propose a novel method for adapting a classifier to a new unseen target user without the need of acquiring additional labeled samples. We prove the effectiveness of this method in the context of facial expression analysis showing that our method achieves higher or comparable performance to the state of the art, at a drastically reduced time cost.

Gli stili APA, Harvard, Vancouver, ISO e altri

50

Zen, Gloria. "Understanding Visual Information: from Unsupervised Discovery to Minimal Effort Domain Adaptation". Doctoral thesis, University of Trento, 2015. http://eprints-phd.biblio.unitn.it/1443/1/GZen_final_thesis.pdf.

Testo completo

Abstract (sommario):

Visual data interpretation is a fascinating problem which has received an increasing attention in the last decades. Reasons for this growing trend can be found within multiple interconnected factors, such as the exponential growth of visual data (e.g. images and videos) availability, the consequent demand for an automatic way to interpret these data and the increase of computational power. In a supervised machine learning approach, a large effort within the research community has been devoted to the collection of training samples to be provided to the learning system, resulting in the generation of very large scale datasets. This has lead to remarkable performance advances in tasks such as scene recognition or object detection, however, at a considerable high cost in terms of human labeling effort. In light of the labeling cost issue, together with the dataset bias one, another significant research direction was headed towards developing methods for learning without or with a limited amount of training data, by leveraging instead on data properties like intrinsic redundancy, time constancy or commonalities shared among different domains. Our work is in line with this last type of approach. In particular, by covering different case scenarios - from dynamic crowded scenes to facial expression analysis - we propose a novel approach to overcome some of the state-of-the-art limitations. Based on the renowned bag of words (BoW) approach, we propose a novel method which achieves higher performances in tasks such as learning typical patterns of behaviors and anomalies discovery from complex scenes, by considering the similarity among visual words in the learning phase. We also show that including sparsity constraints can help dealing with noise which is intrinsic to low level cues extracted from complex dynamic scenes. Facing the so called dataset bias issue, we propose a novel method for adapting a classifier to a new unseen target user without the need of acquiring additional labeled samples. We prove the effectiveness of this method in the context of facial expression analysis showing that our method achieves higher or comparable performance to the state of the art, at a drastically reduced time cost.

Gli stili APA, Harvard, Vancouver, ISO e altri

Tesi sul tema "Adaptation de domaines"

Cita una fonte nei formati APA, MLA, Chicago, Harvard e in molti altri stili