Algoritmos de aprendizado semissupervisionado aprendem a partir de uma combinação de dados rotulados e não rotulados. Assim, eles podem ser aplicados em domínios em que poucos exemplos rotulados e uma vasta quantidade de exemplos não rotulados estão disponíveis. Além disso, os algoritmos semissupervisionados podem atingir um desempenho superior aos algoritmos supervisionados treinados nos mesmos poucos exemplos rotulados. Uma poderosa abordagem ao aprendizado semissupervisionado, denominada aprendizado multidescrição, pode ser usada sempre que os exemplos de treinamento são descritos por dois ou mais conjuntos de atributos disjuntos. A classificação de textos é um domínio de aplicação no qual algoritmos semissupervisionados vêm obtendo sucesso. No entanto, o aprendizado semissupervisionado multidescrição ainda não foi bem explorado nesse domínio dadas as diversas maneiras possíveis de se descrever bases de textos. O objetivo neste trabalho é analisar o desempenho de algoritmos semissupervisionados multidescrição na classificação de textos, usando unigramas e bigramas para compor duas descrições distintas de documentos textuais. Assim, é considerado inicialmente o difundido algoritmo multidescrição CO-TRAINING, para o qual são propostas modificações a fim de se tratar o problema dos pontos de contenção. É também proposto o algoritmo COAL, o qual pode melhorar ainda mais o algoritmo CO-TRAINING pela incorporação de aprendizado ativo como uma maneira de tratar pontos de contenção. Uma ampla avaliação experimental desses algoritmos foi conduzida em bases de textos reais. Os resultados mostram que o algoritmo COAL, usando unigramas como uma descrição das bases textuais e bigramas como uma outra descrição, atinge um desempenho significativamente melhor que um algoritmo semissupervisionado monodescrição. Levando em consideração os bons resultados obtidos por COAL, conclui-se que o uso de unigramas e bigramas como duas descrições distintas de bases de textos pode ser bastante compensador / Semi-supervised learning algorithms learn from a combination of both labeled and unlabeled data. Thus, they can be applied in domains where few labeled examples and a vast amount of unlabeled examples are available. Furthermore, semi-supervised learning algorithms may achieve a better performance than supervised learning algorithms trained on the same few labeled examples. A powerful approach to semi-supervised learning, called multi-view learning, can be used whenever the training examples are described by two or more disjoint sets of attributes. Text classification is a domain in which semi-supervised learning algorithms have shown some success. However, multi-view semi-supervised learning has not yet been well explored in this domain despite the possibility of describing textual documents in a myriad of ways. The aim of this work is to analyze the effectiveness of multi-view semi-supervised learning in text classification using unigrams and bigrams as two distinct descriptions of text documents. To this end, we initially consider the widely adopted CO-TRAINING multi-view algorithm and propose some modifications to it in order to deal with the problem of contention points. We also propose the COAL algorithm, which further improves CO-TRAINING by incorporating active learning as a way of dealing with contention points. A thorough experimental evaluation of these algorithms was conducted on real text data sets. The results show that the COAL algorithm, using unigrams as one description of text documents and bigrams as another description, achieves significantly better performance than a single-view semi-supervised algorithm. Taking into account the good results obtained by COAL, we conclude that the use of unigrams and bigrams as two distinct descriptions of text documents can be very effective

Dans le domaine de l'apprentissage machine, les caractéristiques des données varient généralement dans l'espace des entrées : la distribution globale pourrait être multimodale et contenir des non-linéarités. Afin d'obtenir de bonnes performances, l'algorithme d'apprentissage devrait alors être capable de capturer et de s'adapter à ces changements. Même si les modèles linéaires ne parviennent pas à décrire des distributions complexes, ils sont réputés pour leur passage à l'échelle, en entraînement et en test, aux grands ensembles de données en termes de nombre d'exemples et de nombre de fonctionnalités. Plusieurs méthodes ont été proposées pour tirer parti du passage à l'échelle et de la simplicité des hypothèses linéaires afin de construire des modèles aux grandes capacités discriminatoires. Ces méthodes améliorent les modèles linéaires, dans le sens où elles renforcent leur expressivité grâce à différentes techniques. Cette thèse porte sur l'amélioration des approches d'apprentissage locales, une famille de techniques qui infère des modèles en capturant les caractéristiques locales de l'espace dans lequel les observations sont intégrées.L'hypothèse fondatrice de ces techniques est que le modèle appris doit se comporter de manière cohérente sur des exemples qui sont proches, ce qui implique que ses résultats doivent aussi changer de façon continue dans l'espace des entrées. La localité peut être définie sur la base de critères spatiaux (par exemple, la proximité en fonction d'une métrique choisie) ou d'autres relations fournies, telles que l'association à la même catégorie d'exemples ou un attribut commun. On sait que les approches locales d'apprentissage sont efficaces pour capturer des distributions complexes de données, évitant de recourir à la sélection d'un modèle spécifique pour la tâche. Cependant, les techniques de pointe souffrent de trois inconvénients majeurs :ils mémorisent facilement l'ensemble d'entraînement, ce qui se traduit par des performances médiocres sur de nouvelles données ; leurs prédictions manquent de continuité dans des endroits particuliers de l'espace ; elles évoluent mal avec la taille des ensembles des données. Les contributions de cette thèse examinent les problèmes susmentionnés dans deux directions : nous proposons d'introduire des informations secondaires dans la formulation du problème pour renforcer la continuité de la prédiction et atténuer le phénomène de la mémorisation ; nous fournissons une nouvelle représentation de l'ensemble de données qui tient compte de ses spécificités locales et améliore son évolutivité. Des études approfondies sont menées pour mettre en évidence l'efficacité de ces contributions pour confirmer le bien-fondé de leurs intuitions. Nous étudions empiriquement les performances des méthodes proposées tant sur des jeux de données synthétiques que sur des tâches réelles, en termes de précision et de temps d'exécution, et les comparons aux résultats de l'état de l'art. Nous analysons également nos approches d'un point de vue théorique, en étudiant leurs complexités de calcul et de mémoire et en dérivant des bornes de généralisation serrées. / In Machine Learning field, data characteristics usually vary over the space: the overall distribution might be multi-modal and contain non-linearities.In order to achieve good performance, the learning algorithm should then be able to capture and adapt to these changes. Even though linear models fail to describe complex distributions, they are renowned for their scalability, at training and at testing, to datasets big in terms of number of examples and of number of features. Several methods have been proposed to take advantage of the scalability and the simplicity of linear hypotheses to build models with great discriminatory capabilities. These methods empower linear models, in the sense that they enhance their expressive power through different techniques. This dissertation focuses on enhancing local learning approaches, a family of techniques that infers models by capturing the local characteristics of the space in which the observations are embedded. The founding assumption of these techniques is that the learned model should behave consistently on examples that are close, implying that its results should also change smoothly over the space. The locality can be defined on spatial criteria (e.g. closeness according to a selected metric) or other provided relations, such as the association to the same category of examples or a shared attribute. Local learning approaches are known to be effective in capturing complex distributions of the data, avoiding to resort to selecting a model specific for the task. However, state of the art techniques suffer from three major drawbacks: they easily memorize the training set, resulting in poor performance on unseen data; their predictions lack of smoothness in particular locations of the space;they scale poorly with the size of the datasets. The contributions of this dissertation investigate the aforementioned pitfalls in two directions: we propose to introduce side information in the problem formulation to enforce smoothness in prediction and attenuate the memorization phenomenon; we provide a new representation for the dataset which takes into account its local specificities and improves scalability. Thorough studies are conducted to highlight the effectiveness of the said contributions which confirmed the soundness of their intuitions. We empirically study the performance of the proposed methods both on toy and real tasks, in terms of accuracy and execution time, and compare it to state of the art results. We also analyze our approaches from a theoretical standpoint, by studying their computational and memory complexities and by deriving tight generalization bounds.

近年來諸多學者專家致力於從多視角影像獲取精確的點雲資訊，並藉由點雲資訊進行三維模型重建等研究，然而透過多視角影像求取三維資訊的精確度仍然有待提升，其中萃取影像對應與重建三維資訊方法，是多視角影像重建三維資訊的關鍵核心，決定點雲資訊的形成方式與成效。 本論文中，我們提出了一套新的方法，由多視角影像之間的幾何關係出發，萃取多視角影像對應與重建三維點，可以有效地改善對應點與三維點的精確度。首先，在萃取多視角影像對應的部份，我們以相互支持轉換、動態高斯濾波法與綜合性相似度評估函數，改善補綴面為基礎的比對方法，提高相似度測量值的辨識力與可信度，可從多視角影像中獲得精確的對應點。其次，在重建三維點的部份，我們使用K均值分群演算法與線性內插法發掘潛在的三維點，讓求出的三維點更貼近三維空間真實物體表面，能在多視角影像中獲得更精確的三維點。 實驗結果顯示，採用本研究所提出的方法進行改善後，在對應點精確度的提升上有很好的成效，所獲得的點雲資訊存在數萬個精確的三維點，而且僅有少數的離群點。 / Recently, many researchers pay attentions in obtaining accurate point cloud data from multi-view images and use these data in 3D model reconstruction. However, this accuracy still needs to be improved. Among these researches, the methods in extracting the corresponding points as well as computing the 3D point information are the most critical ones. These methods practically affect the final results of the point cloud data and the 3D models so constructed. In this thesis, we propose new approaches, based on multi-view geometry, to improve the accuracy of corresponding points and 3D points. Mutual support transformation, dynamic Gaussian filtering, and similarity evaluation function were used to improve the patch-based matching methods in multi-view image correspondence. Using these mechanisms, the discrimination ability and reliability of the similarity function and, hence, the accuracy of the extracted corresponding points can be greatly improved. We also used K-mean algorithms and linear interpolations to find the better 3D point candidates. The 3D point so computed will be much closer to the surface of the actual 3D object. Thus, this mechanism will produce highly accurate 3D points. Experimental results show that our mechanism can improve the accuracy of corresponding points as well as the 3D point cloud data. We successfully generated accurate point cloud data that contains tens of thousands 3D points, and, moreover, only has a few outliers.

La classification automatique non supervisée est un problème majeur, aux frontières de multiples communautés issues de l’Intelligence Artificielle, de l’Analyse de Données et des Sciences de la Cognition. Elle vise à formaliser et mécaniser la tâche cognitive de classification, afin de l’automatiser pour la rendre applicable à un grand nombre d’objets (ou individus) à classer. Des visées plus applicatives s’intéressent à l’organisation automatique de grands ensembles d’objets en différents groupes partageant des caractéristiques communes. La présente thèse propose des méthodes de classification non supervisées applicables lorsque plusieurs sources d’informations sont disponibles pour compléter et guider la recherche d’une ou plusieurs classifications des données. Pour la classification non supervisée multi-vues, la première contribution propose un mécanisme de recherche de classifications locales adaptées aux données dans chaque représentation, ainsi qu’un consensus entre celles-ci. Pour la classification semi-supervisée, la seconde contribution propose d’utiliser des connaissances externes sur les données pour guider et améliorer la recherche d’une classification d’objets par un algorithme quelconque de partitionnement de données. Enfin, la troisième et dernière contribution propose un environnement collaboratif permettant d’atteindre au choix les objectifs de consensus et d’alternatives pour la classification d’objets mono-représentés ou multi-représentés. Cette dernière contribution ré-pond ainsi aux différents problèmes de multiplicité des données et des analyses dans le contexte de la classification non supervisée, et propose, au sein d’une même plate-forme unificatrice, une proposition répondant à des problèmes très actifs et actuels en Fouille de Données et en Extraction et Gestion des Connaissances. / Data clustering is a major problem encountered mainly in related fields of Artificial Intelligence, Data Analysis and Cognitive Sciences. This topic is concerned by the production of synthetic tools that are able to transform a mass of information into valuable knowledge. This knowledge extraction is done by grouping a set of objects associated with a set of descriptors such that two objects in a same group are similar or share a same behaviour while two objects from different groups does not. This thesis present a study about some extensions of the classical clustering problem for multi-view data,where each datum can be represented by several sets of descriptors exhibing different behaviours or aspects of it. Our study impose to explore several nearby problems such that semi-supervised clustering, multi-view clustering or collaborative approaches for consensus or alternative clustering. In a first chapter, we propose an algorithm solving the multi-view clustering problem. In the second chapter, we propose a boosting-inspired algorithm and an optimization based algorithm closely related to boosting that allow the integration of external knowledge leading to the improvement of any clustering algorithm. This proposition bring an answer to the semi-supervised clustering problem. In the last chapter, we introduce an unifying framework allowing the discovery even of a set of consensus clustering solution or a set of alternative clustering solutions for mono-view data and or multi-viewdata. Such unifying approach offer a methodology to answer some current and actual hot topic in Data Mining and Knowledge Discovery in Data.

Les technologies de reconstruction multi-vues permettent de réaliser un clone virtuel d'un acteur à partir d'une simple acquisition vidéo réalisée par un ensemble de caméras à partir de multiples points de vue. Cette approche offre de nouvelles opportunités dans le domaine de la composition de scènes hybrides mélangeant les images réelles et virtuelles. Cette thèse a été réalisée dans le cadre du projet RECOVER 3D dont l'objectif est de développer une chaîne de production TV complète, de l'acquisition jusqu'à la diffusion, autour de la reconstruction multi-vues. Cependant la technologie utilisée dans ce contexte est mal adaptée à la reconstruction de scènes dynamiques. En effet, la performance d'un acteur est reproduite sous la forme d'une séquence d'objets 3D statiques qui correspondent aux poses successives du personnage au cours de la capture vidéo. L'objectif de cette thèse est de développer une méthode pour transformer ces séquences de poses en un modèle animé unique. Les travaux de recherches menés dans ce cadre sont répartis en deux étapes principales. La première a pour but de calculer un champ de déplacements qui décrit les mouvements de l'acteur entre deux poses consécutives. La seconde étape consiste à animer un maillage suivant les trajectoires décrites par le champ de mouvements, de manière à le déplacer vers la pose suivante. En répétant ce processus tout au long la séquence, nous parvenons ainsi à reproduire un maillage animé qui adopte les poses successives de l'acteur. Les résultats obtenus montrent que notre méthode peut générer un modèle temporellement cohérent à partir d'une séquence d'enveloppes visuelles. / 4D multi-view reconstruction technologies are more and more used in media production due to their abilities to produce a virtual clone of an actor from a simple video acquisition performed by a set of multi-viewpoint cameras. This approach is a major advance for the composition of animations which mix virtual and real images, and also offers new possibilities for the rendering of such complex hybrid scenes. The work described in this thesis takes parts in the RECOVER 3D project which aims at developing an innovative industrial framework for TV production, based on multi-view reconstruction, from studio acquisition to broadcasting. The major drawback of the methods used in this context is that they are not adapted to the reconstruction of dynamic scenes. The output are time series which describe the successive poses of the actor, figured as a sequence of static objects. The goal of this thesis is to transform these initial results into a dynamic 3D object where the actor is figured as an animated character. The research detailed in this manuscript presents two main contributions. The first one is centered on the computation of a motion flow which represents the displacements occurring in the reconstructed scene between two consecutive poses. The second one presents a mesh animation process that leads to the animation of a 3D model from one pose to another, following the motion flow. This two-step operation is repeated throughout the entire pose sequence to finally obtain a single animated mesh that matches the evolving shape of the reconstructed actor. Results show that our method is able to produce a temporally consistent mesh animation from various sequences of visual hulls.

Dans cette thèse, nous traitons des problèmes de la détection des personnes et de l'estimation de leurs poses dans la Salle Opératoire (SO), deux éléments clés pour le développement d'applications d'assistance chirurgicale. Nous percevons la salle grâce à des caméras RGB-D qui fournissent des informations visuelles complémentaires sur la scène. Ces informations permettent de développer des méthodes mieux adaptées aux difficultés propres aux SO, comme l'encombrement, les surfaces sans texture et les occlusions. Nous présentons des nouvelles approches qui tirent profit des informations temporelles, de profondeur et des vues multiples afin de construire des modèles robustes pour la détection des personnes et de leurs poses. Une évaluation est effectuée sur plusieurs jeux de données complexes enregistrés dans des salles opératoires avec une ou plusieurs caméras. Les résultats obtenus sont très prometteurs et montrent que nos approches surpassent les méthodes de l'état de l'art sur ces données cliniques. / In this thesis, we address the two problems of person detection and pose estimation in Operating Rooms (ORs), which are key ingredients in the development of surgical assistance applications. We perceive the OR using compact RGB-D cameras that can be conveniently integrated in the room. These sensors provide complementary information about the scene, which enables us to develop methods that can cope with numerous challenges present in the OR, e.g. clutter, textureless surfaces and occlusions. We present novel part-based approaches that take advantage of depth, multi-view and temporal information to construct robust human detection and pose estimation models. Evaluation is performed on new single- and multi-view datasets recorded in operating rooms. We demonstrate very promising results and show that our approaches outperform state-of-the-art methods on this challenging data acquired during real surgeries.

Cette thèse a pour objectif la conception de méthodes pour la reconnaissance automatique des activités chirurgicales. Cette reconnaissance est un élément clé pour le développement de systèmes réactifs au contexte clinique et pour des applications comme l’assistance automatique lors de chirurgies complexes. Nous abordons ce problème en utilisant des méthodes de Vision puisque l’utilisation de caméras permet de percevoir l’environnement sans perturber la chirurgie. Deux types de vidéos sont utilisées : des vidéos laparoscopiques et des vidéos multi-vues RGBD. Nous avons d’abord étudié les résultats obtenus avec les méthodes de l’état de l’art, puis nous avons proposé des nouvelles approches basées sur le « Deep learning ». Nous avons aussi généré de larges jeux de données constitués d’enregistrements de chirurgies. Les résultats montrent que nos méthodes permettent d’obtenir des meilleures performances pour la reconnaissance automatique d’activités chirurgicales que l’état de l’art. / The main objective of this thesis is to address the problem of activity recognition in the operating room (OR). Activity recognition is an essential component in the development of context-aware systems, which will allow various applications, such as automated assistance during difficult procedures. Here, we focus on vision-based approaches since cameras are a common source of information to observe the OR without disrupting the surgical workflow. Specifically, we propose to use two complementary video types: laparoscopic and OR-scene RGBD videos. We investigate how state-of-the-art computer vision approaches perform on these videos and propose novel approaches, consisting of deep learning approaches, to carry out the tasks. To evaluate our proposed approaches, we generate large datasets of recordings of real surgeries. The results demonstrate that the proposed approaches outperform the state-of-the-art methods in performing surgical activity recognition on these new datasets.

De nos jours, dans plusieurs domaines, tels que la bio-informatique ou le multimédia, les données peuvent être représentées par plusieurs ensembles d'attributs, appelés des vues. Pour une tâche de classification donnée, nous distinguons deux types de vues : les vues fortes sont celles adaptées à la tâche, les vues faibles sont adaptées à une (petite) partie de la tâche ; en classification multi-classes, chaque vue peut s'avérer forte pour reconnaître une classe, et faible pour reconnaître d’autres classes : une telle vue est dite déséquilibrée. Les travaux présentés dans cette thèse s'inscrivent dans le cadre de l'apprentissage supervisé et ont pour but de traiter les questions d'apprentissage multi-vue dans le cas des vues fortes, faibles et déséquilibrées. La première contribution de cette thèse est un algorithme d'apprentissage multi-vues théoriquement fondé sur le cadre de boosting multi-classes utilisé par AdaBoost.MM. La seconde partie de cette thèse concerne la mise en place d'un cadre général pour les méthodes d'apprentissage de classes déséquilibrées (certaines classes sont plus représentées que les autres). Dans la troisième partie, nous traitons le problème des vues déséquilibrées en combinant notre approche des classes déséquilibrées et la coopération entre les vues mise en place pour appréhender la classification multi-vues. Afin de tester les méthodes sur des données réelles, nous nous intéressons au problème de classification d'appels téléphoniques, qui a fait l'objet du projet ANR DECODA. Ainsi chaque partie traite différentes facettes du problème. / Nowadays, in many fields, such as bioinformatics or multimedia, data may be described using different sets of features, also called views. For a given classification task, we distinguish two types of views:strong views, which are suited for the task, and weak views suited for a (small) part of the task; in multi-class learning, a view can be strong with respect to some (few) classes and weak for the rest of the classes: these are imbalanced views. The works presented in this thesis fall in the supervised learning setting and their aim is to address the problem of multi-view learning under strong, weak and imbalanced views, regrouped under the notion of uneven views. The first contribution of this thesis is a multi-view learning algorithm based on the same framework as AdaBoost.MM. The second part of this thesis proposes a unifying framework for imbalanced classes supervised methods (some of the classes are more represented than others). In the third part of this thesis, we tackle the uneven views problem through the combination of the imbalanced classes framework and the between-views cooperation used to take advantage of the multiple views. In order to test the proposed methods on real-world data, we consider the task of phone calls classifications, which constitutes the subject of the ANR DECODA project. Each part of this thesis deals with different aspects of the problem.

Algoritmos de aprendizado semissupervisionado aprendem a partir de uma combinação de dados rotulados e não rotulados. Assim, eles podem ser aplicados em domínios em que poucos exemplos rotulados e uma vasta quantidade de exemplos não rotulados estão disponíveis. Além disso, os algoritmos semissupervisionados podem atingir um desempenho superior aos algoritmos supervisionados treinados nos mesmos poucos exemplos rotulados. Uma poderosa abordagem ao aprendizado semissupervisionado, denominada aprendizado multidescrição, pode ser usada sempre que os exemplos de treinamento são descritos por dois ou mais conjuntos de atributos disjuntos. A classificação de textos é um domínio de aplicação no qual algoritmos semissupervisionados vêm obtendo sucesso. No entanto, o aprendizado semissupervisionado multidescrição ainda não foi bem explorado nesse domínio dadas as diversas maneiras possíveis de se descrever bases de textos. O objetivo neste trabalho é analisar o desempenho de algoritmos semissupervisionados multidescrição na classificação de textos, usando unigramas e bigramas para compor duas descrições distintas de documentos textuais. Assim, é considerado inicialmente o difundido algoritmo multidescrição CO-TRAINING, para o qual são propostas modificações a fim de se tratar o problema dos pontos de contenção. É também proposto o algoritmo COAL, o qual pode melhorar ainda mais o algoritmo CO-TRAINING pela incorporação de aprendizado ativo como uma maneira de tratar pontos de contenção. Uma ampla avaliação experimental desses algoritmos foi conduzida em bases de textos reais. Os resultados mostram que o algoritmo COAL, usando unigramas como uma descrição das bases textuais e bigramas como uma outra descrição, atinge um desempenho significativamente melhor que um algoritmo semissupervisionado monodescrição. Levando em consideração os bons resultados obtidos por COAL, conclui-se que o uso de unigramas e bigramas como duas descrições distintas de bases de textos pode ser bastante compensador / Semi-supervised learning algorithms learn from a combination of both labeled and unlabeled data. Thus, they can be applied in domains where few labeled examples and a vast amount of unlabeled examples are available. Furthermore, semi-supervised learning algorithms may achieve a better performance than supervised learning algorithms trained on the same few labeled examples. A powerful approach to semi-supervised learning, called multi-view learning, can be used whenever the training examples are described by two or more disjoint sets of attributes. Text classification is a domain in which semi-supervised learning algorithms have shown some success. However, multi-view semi-supervised learning has not yet been well explored in this domain despite the possibility of describing textual documents in a myriad of ways. The aim of this work is to analyze the effectiveness of multi-view semi-supervised learning in text classification using unigrams and bigrams as two distinct descriptions of text documents. To this end, we initially consider the widely adopted CO-TRAINING multi-view algorithm and propose some modifications to it in order to deal with the problem of contention points. We also propose the COAL algorithm, which further improves CO-TRAINING by incorporating active learning as a way of dealing with contention points. A thorough experimental evaluation of these algorithms was conducted on real text data sets. The results show that the COAL algorithm, using unigrams as one description of text documents and bigrams as another description, achieves significantly better performance than a single-view semi-supervised algorithm. Taking into account the good results obtained by COAL, we conclude that the use of unigrams and bigrams as two distinct descriptions of text documents can be very effective

Em Aprendizado de Máquina, a abordagem supervisionada normalmente necessita de um número significativo de exemplos de treinamento para a indução de classificadores precisos. Entretanto, a rotulação de dados é freqüentemente realizada manualmente, o que torna esse processo demorado e caro. Por outro lado, exemplos não-rotulados são facilmente obtidos se comparados a exemplos rotulados. Isso é particularmente verdade para tarefas de classificação de textos que envolvem fontes de dados on-line tais como páginas de internet, email e artigos científicos. A classificação de textos tem grande importância dado o grande volume de textos disponível on-line. Aprendizado semi-supervisionado, uma área de pesquisa relativamente nova em Aprendizado de Máquina, representa a junção do aprendizado supervisionado e não-supervisionado, e tem o potencial de reduzir a necessidade de dados rotulados quando somente um pequeno conjunto de exemplos rotulados está disponível. Este trabalho descreve o algoritmo de aprendizado semi-supervisionado co-training, que necessita de duas descrições de cada exemplo. Deve ser observado que as duas descrições necessárias para co-training podem ser facilmente obtidas de documentos textuais por meio de pré-processamento. Neste trabalho, várias extensões do algoritmo co-training foram implementadas. Ainda mais, foi implementado um ambiente computacional para o pré-processamento de textos, denominado PreTexT, com o objetivo de utilizar co-training em problemas de classificação de textos. Os resultados experimentais foram obtidos utilizando três conjuntos de dados. Dois conjuntos de dados estão relacionados com classificação de textos e o outro com classificação de páginas de internet. Os resultados, que variam de excelentes a ruins, mostram que co-training, similarmente a outros algoritmos de aprendizado semi-supervisionado, é afetado de maneira bastante complexa pelos diferentes aspectos na indução dos modelos. / In Machine Learning, the supervised approach usually requires a large number of labeled training examples to learn accurately. However, labeling is often manually performed, making this process costly and time-consuming. By contrast, unlabeled examples are often inexpensive and easier to obtain than labeled examples. This is especially true for text classification tasks involving on-line data sources, such as web pages, email and scientific papers. Text classification is of great practical importance today given the massive volume of online text available. Semi-supervised learning, a relatively new area in Machine Learning, represents a blend of supervised and unsupervised learning, and has the potential of reducing the need of expensive labeled data whenever only a small set of labeled examples is available. This work describes the semi-supervised learning algorithm co-training, which requires a partitioned description of each example into two distinct views. It should be observed that the two different views required by co-training can be easily obtained from textual documents through pre-processing. In this works, several extensions of co-training algorithm have been implemented. Furthermore, we have also implemented a computational environment for text pre-processing, called PreTexT, in order to apply the co-training algorithm to text classification problems. Experimental results using co-training on three data sets are described. Two data sets are related to text classification and the other one to web-page classification. Results, which range from excellent to poor, show that co-training, similarly to other semi-supervised learning algorithms, is affected by modelling assumptions in a rather complicated way.