1,721,112 research outputs found
VideoSense at TRECVID 2011 : Semantic indexing from light similarity functions-based domain adaptation with stacking
No full text
Going Beyond Counting First Authors in Author Co-citation Analysis
Get PDFThe present study examines one of the fundamental aspects of author co-citation analysis (ACA) - the way co-citation
counts are defined. Co-citation counting provides the data on which all subsequent statistical analyses and mappings
are based, and we compare ACA results based on two different types of co-citation counting - the traditional type that
only counts the first one among a cited work's authors on the one hand and a non-traditional type that takes into
account the first 5 authors of a cited work on the other hand. Results indicate that the picture produced through this non-traditional author co-citation counting contains more coherent author groups and is therefore considerably clearer. However, this picture represents fewer specialties in the research field being studied than that produced through the traditional first-author co-citation counting when the same number of top-ranked authors is selected and analyzed. Reasons for these effects are discussed
Variations on the Author
Get PDF“Variations on the Author” discusses two of Eduardo Coutinho’s recent films (Um Dia na Vida, from 2010, and Últimas Conversas, posthumously released in 2015) and their contribution to the general question of documentary authorship. The director’s filmography is characterized by a consistent yet self-effacing form of authorial self-inscription: Coutinho often features as an interviewer that rather than express opinions propels discourses; an interviewer that is good at listening. This mode of self-inscription characterizes him as an author who is not expressive but who is nonetheless markedly present on the screen. In Um Dia na Vida, however, Coutinho is completely absent form the image, while Últimas Conversas, on the contrary, includes a confessional prologue that moves the director from the margins to the center of his films. This article examines the ways in which these works stand out in the filmography of a director who offers new insights into the notion of cinematic authorship
Appropriate Similarity Measures for Author Cocitation Analysis
Get PDFWe provide a number of new insights into the methodological discussion about author cocitation analysis. We first argue that the use of the Pearson correlation for measuring the similarity between authors’ cocitation profiles is not very satisfactory. We then discuss what kind of similarity measures may be used as an alternative to the Pearson correlation. We consider three similarity measures in particular. One is the well-known cosine. The other two similarity measures have not been used before in the bibliometric literature. Finally, we show by means of an example that our findings have a high practical relevance.information science;Pearson correlation;cosine;similarity measure;author cocitation analysis
Apprentissage de vote de majorité pour la classification supervisée et l'adaptation de domaine : approches PAC-Bayésiennes et combinaison de similarités
No full textNowadays, due to the expansion of the web a plenty of data are available and many applications need to make use of supervised machine learning methods able to take into account different information sources. For instance, for multimedia semantic indexing applications, one have to efficiently take advantage of information about color, textual, texture or sound sources of the document. Most of the existing methods try to combine these multimodal informations, either by directly fusionning the descriptors or by combining similarities or classifiers, in order to produce a classification model more reliable for the considered task. Usually, these multimodal facets imply two main issues. On the one hand, one have to be able to correctly make use of all the a priori information available. On the other hand, the data, on which the model will be applied, does not come from the same probability distribution than the data used during the learning step. In this context, we have to adapt the model on new data, which is known as domain adaptation. In this thesis, we propose several theoretically-founded contributions for tackle these issues. A first serie of contributions studies the problem of learning a weighted majority vote over a set of voters in a supervised classification setting.These results fall within the context of the PAC-Bayesian theory allowing to derive generalization abilities for such a vote by assuming an a priori on the relevance of the voters. Our first contribution aims at extending a recent algorithm, MinCq, minimizing a bound over the error of the majority vote in binary classification. This extension can take into account an a priori belief on the performances of the voters. This belief is expressed as an aligned distribution. We illustrate its usefulness for combining nearest neighbor classifiers, and for classifier fusion on a multimedia semantic indexing task. Then, we propose a theoretical contribution for multiclass classification tasks. Our approach is based on an original PAC-Bayesian analysis considering the operator norm of the confusion matrix as an error measure. Our second series of contributions relates to domain adaptation. In this situation we present our third result for combining similarities in order to infer a representation space for moving closer the learning distribution and the testing distribution. This contribution is based on the theory of learning from (epsilon,gamma,tau)-good similarity functions and is justified by the minimization of an usual bound in domain adaptation. For our last contribution, we propose the first PAC-Bayesian analysis for domain adaptation. This analysis is based on a consistent divergence measure between distributions allowing us to derive a generalization bound for learning majority votes in binary classification. Moreover, we propose a first algorithm specialized to linear classifiers and able to directly minimize our bound.De nos jours, avec l'expansion d'Internet, l'abondance et la diversité des données accessibles qui en résulte, de nombreuses applications requièrent l'utilisation de méthodes d'apprentissage automatique supervisé capables de prendre en considération différentes sources d'informations. Par exemple, pour des applications relevant de l'indexation sémantique de documents multimédia, il s'agit de pouvoir efficacement tirer bénéfice d'informations liées à la couleur, au texte, à la texture ou au son des documents à traiter. La plupart des méthodes existantes proposent de combiner ces informations multimodales, soit en fusionnant directement les descriptions, soit en combinant des similarités ou des classifieurs, avec pour objectif de construire un modèle de classification automatique plus fiable pour la tâche visée. Ces aspects multimodaux induisent généralement deux types de difficultés. D'une part, il faut être capable d'utiliser au mieux toute l'information a priori disponible sur les objets à combiner. D'autre part, les données sur lesquelles le modèle doit être appliqué ne suivent nécessairement pas la même distribution de probabilité que les données utilisées lors de la phase d'apprentissage. Dans ce contexte, il faut être à même d'adapter le modèle à de nouvelles données, ce qui relève de l'adaptation de domaine. Dans cette thèse, nous proposons plusieurs contributions fondées théoriquement et répondant à ces problématiques. Une première série de contributions s'intéresse à l'apprentissage de votes de majorité pondérés sur un ensemble de votants dans le cadre de la classification supervisée. Ces contributions s'inscrivent dans le contexte de la théorie PAC-Bayésienne permettant d'étudier les capacités en généralisation de tels votes de majorité en supposant un a priori sur la pertinence des votants. Notre première contribution vise à étendre un algorithme récent, MinCq, minimisant une borne sur l'erreur du vote de majorité en classification binaire. Cette extension permet de prendre en compte une connaissance a priori sur les performances des votants à combiner sous la forme d'une distribution alignée. Nous illustrons son intérêt dans une optique de combinaison de classifieurs de type plus proches voisins, puis dans une perspective de fusion de classifieurs pour l'indexation sémantique de documents multimédia. Nous proposons ensuite une contribution théorique pour des problèmes de classification multiclasse. Cette approche repose sur une analyse PAC-Bayésienne originale en considérant la norme opérateur de la matrice de confusion comme mesure de risque. Notre seconde série de contributions concerne la problématique de l'adaptation de domaine. Dans cette situation, nous présentons notre troisième apport visant à combiner des similarités permettant d'inférer un espace de représentation de manière à rapprocher les distributions des données d'apprentissage et des données à traiter. Cette contribution se base sur la théorie des fonctions de similarités (epsilon,gamma,tau)-bonnes et se justifie par la minimisation d'une borne classique en adaptation de domaine. Pour notre quatrième et dernière contribution, nous proposons la première analyse PAC-Bayésienne appropriée à l'adaptation de domaine. Cette analyse se base sur une mesure consistante de divergence entre distributions permettant de dériver une borne en généralisation pour l'apprentissage de votes de majorité en classification binaire. Elle nous permet également de proposer un algorithme adapté aux classifieurs linéaires capable de minimiser cette borne de manière directe
Avancées en théorie PAC-Bayésienne : de bornes en généralisation à des algorithmes d'apprentissage supervisé et de transfert
No full textAdaptation de domaine de vote de majorité par auto-étiquetage non itératif
No full textNational audienceEn apprentissage automatique, nous parlons d'adaptation de domaine lorsque les données de test (cibles) et d'apprentissage (sources) sont générées selon différentes distributions. Nous devons donc développer des algorithmes de classification capables de s'adapter à une nouvelle distribution, pour laquelle aucune information sur les étiquettes n'est disponible. Nous attaquons cette problématique sous l'angle de l'approche PAC-Bayésienne qui se focalise sur l'apprentissage de modèles définis comme des votes de majorité sur un ensemble de fonctions. Dans ce contexte, nous introduisons PV-MinCq une version adaptative de l'algorithme (non adaptatif) MinCq. PV-MinCq suit le principe suivant. Nous transférons les étiquettes sources aux points cibles proches pour ensuite appliquer MinCq sur l'échantillon cible ''auto-étiqueté'' (justifié par une borne théorique). Plus précisément, nous définissons un auto-étiquetage non itératif qui se focalise dans les régions où les distributions marginales source et cible sont les plus similaires. Dans un second temps, nous étudions l'influence de notre auto-étiquetage pour en déduire une procédure de validation des hyperparamètres. Finalement, notre approche montre des résultats empiriques prometteurs
Domain Adaptation of Majority Votes via Perturbed Variation-based Label Transfer
No full textWe tackle the PAC-Bayesian Domain Adaptation (DA) problem. This arrives when one desires to learn, from a source distribution, a good weighted majority vote (over a set of classifiers) on a different target distribution. In this context, the disagreement between classifiers is known crucial to control. In non-DA supervised setting, a theoretical bound - the C-bound - involves this disagreement and leads to a majority vote learning algorithm: MinCq. In this work, we extend MinCq to DA by taking advantage of an elegant divergence between distribution called the Perturbed Varation (PV). Firstly, justified by a new formulation of the C-bound, we provide to MinCq a target sample labeled thanks to a PV-based self-labeling focused on regions where the source and target marginal distributions are closer. Secondly, we propose an original process for tuning the hyperparameters. Our framework shows very promising results on a toy problem
Domain adaptation of weighted majority votes via perturbed variation-based self-labeling
No full textThe published version is available here: http://www.sciencedirect.com/science/article/pii/S0167865514002736International audienceIn machine learning, the domain adaptation problem arrives when the test (target) and the train (source) data are generated from different distributions. A key applied issue is thus the design of algorithms able to generalize on a new distribution, for which we have no label information. We focus on learning classification models defined as a weighted majority vote over a set of real-val ued functions. In this context, Germain et al. (2013) have shown that a measure of disagreement between these functions is crucial to control. The core of this measure is a theoretical bound--the C-bound (Lacasse et al., 2007)--which involves the disagreement and leads to a well performing majority vote learning algorithm in usual non-adaptative supervised setting: MinCq. In this work, we propose a framework to extend MinCq to a domain adaptation scenario. This procedure takes advantage of the recent perturbed variation divergence between distributions proposed by Harel and Mannor (2012). Justified by a theoretical bound on the target risk of the vote, we provide to MinCq a target sample labeled thanks to a perturbed variation-based self-labeling focused on the regions where the source and target marginals appear similar. We also study the influence of our self-labeling, from which we deduce an original process for tuning the hyperparameters. Finally, our framework called PV-MinCq shows very promising results on a rotation and translation synthetic problem
- …
