Éditeur(s) : HAL CCSD Presses Universitaires de Louvain (distribution I6DOC/CIACO)
Résumé : In French. 10 pages, 5 figures, LaTeX 2e using EPSF and custom package taln2006.sty (designed by Pierre Zweigenbaum, ATALA). Proceedings of the 13th annual French-speaking conference on Natural Language Processing: `Traitement Automatique des Langues Naturelles' (TALN 2006), Louvain (Leuven), Belgium, 10-13 April 2003
National audience
We propose a theoretical framework within which information on the vocabulary of a given corpus can be inferred on the basis of statistical information gathered on that corpus. Inferences can be made on the categories of the words in the vocabulary, and on their syntactical properties within particular languages. Based on the same statistical data, it is possible to build matrices of syntagmatic similarity (bigram transition matrices) or paradigmatic similarity (probability for any pair of words to share common contexts). When clustered with respect to their syntagmatic similarity, words tend to group into sublanguage vocabularies, and when clustered with respect to their paradigmatic similarity, into syntactic or semantic classes. Experiments have explored the first of these two possibilities. Their results are interpreted in the frame of a Markov chain modelling of the corpus' generative processe(s): we show that the results of a spectral analysis of the transition matrix can be interpreted as probability distributions of words within clusters. This method yields a soft clustering of the vocabulary into sublanguages which contribute to the generation of heterogeneous corpora. As an application, we show how multilingual texts can be visually segmented into linguistically homogeneous segments. Our method is specifically useful in the case of related languages which happened to be mixed in corpora.
Verbum ex machina : Actes de la 13ème conférence annuelle sur le Traitement Automatique des Langues Naturelles
Louvain (Leuven), Belgium

hal-00327607
https://hal.archives-ouvertes.fr/hal-00327607
ARXIV : 0810.1212

Éditeur(s) : HAL CCSD IOS Press (Amsterdam)
Résumé : 6 pages, 7 embedded figures, LaTeX 2e using the ecai2006.cls document class and the algorithm2e.sty style file (+ standard packages like epsfig, amsmath, amssymb, amsfonts...). Extends the short version contained in the ECAI 2006 proceedings.
International audience
Without prior knowledge, distinguishing different languages may be a hard task, especially when their borders are permeable. We develop an extension of spectral clustering -- a powerful unsupervised classification toolbox -- that is shown to resolve accurately the task of soft language distinction. At the heart of our approach, we replace the usual hard membership assignment of spectral clustering by a soft, probabilistic assignment, which also presents the advantage to bypass a well-known complexity bottleneck of the method. Furthermore, our approach relies on a novel, convenient construction of a Markov chain out of a corpus. Extensive experiments with a readily available system clearly display the potential of the method, which brings a visually appealing soft distinction of languages that may define altogether a whole corpus.
ECAI 2006: 17th European Conference on Artificial Intelligence
Riva del Garda, Italy

hal-00327782
https://hal.archives-ouvertes.fr/hal-00327782
ARXIV : 0810.1261

Éditeur(s) : HAL CCSD Elsevier
Résumé : International audience
Voting rules relying on k-nearest neighbors (k-NN) are an effective tool in countless many machine learning techniques. Thanks to its simplicity, k-NN classification is very attractive to practitioners, as it enables very good performances in several practical applications. However, it suffers from various drawbacks, like sensitivity to "noisy" instances and poor generalization properties when dealing with sparse high-dimensional data. In this paper, we tackle the k-NN classification problem at its core by providing a novel k-NN boosting approach. Namely, we propose a supervised learning algorithm, called Universal Nearest Neighbors (UNN), that induces a leveraged k-NN rule by globally minimizing a surrogate risk upper bounding the empirical misclassification rate over training data. Interestingly, this surrogate risk can be arbitrary chosen from a class of Bregman loss functions, including the familiar exponential, logistic and squared losses. Furthermore, we show that UNN allows to efficiently filter a dataset of instances by keeping only a small fraction of data. Experimental results on the synthetic Ripley's dataset show that such a filtering strategy is able to reject "noisy" examples, and yields a classification error close to the optimal Bayes error. Experiments on standard UCI datasets show significant improvements over the current state of the art.
ISSN: 0925-2312

hal-00664462
https://hal.inria.fr/hal-00664462

Éditeur(s) : HAL CCSD IEEE
Résumé : International audience
Recent works display that large scale image classification problems rule out computationally demanding methods. On such problems, simple approaches like k-NN are affordable contenders, with still room space for statistical improvements under the algorithmic constraints. A recent work showed how to leverage k-NN to yield a formal boosting algorithm. This method, however, has numerical issues that make it not suited for large scale problems. We propose here an Adaptive Newton-Raphson scheme to leverage k-NN, N³, which does not suffer these issues. We show that it is a boosting algorithm, with several key algorithmic and statistical properties. In particular, it may be sufficient to boost a subsample to reach desired bounds for the loss at hand in the boosting framework. Experiments are provided on the SUN, and Caltech databases. They confirm that boosting a subsample -- sometimes containing few examples only -- is sufficient to reach the convergence regime of N³. Under such conditions, N³ challenges the accuracy of contenders with lower computational cost and lower memory requirement.
IEEE International Workshop on Machine Learning for Signal Processing
Southampton, United Kingdom

hal-00959125
https://hal.archives-ouvertes.fr/hal-00959125
https://hal.archives-ouvertes.fr/hal-00959125/document
https://hal.archives-ouvertes.fr/hal-00959125/file/BNNB_mlsp2013.pdf

Éditeur(s) : HAL CCSD
Résumé : International audience
The k-nearest neighbors (k-NN) classification rule is still an essential tool for computer vision applications, such as scene recognition. However, k-NN still features some major drawbacks, which mainly reside in the uniform voting among the nearest prototypes in the feature space. In this paper, we propose a new method that is able to learn the "relevance" of prototypes, thus classifying test data using a weighted k-NN rule. In particular, our algorithm, called Multi-class Leveraged k-nearest neighbor (MLNN), learns the prototype weights in a boosting framework, by minimizing a surrogate exponential risk over training data. We propose two main contributions for improving computational speed and accuracy. On the one hand, we implement learning in an inherently multiclass way, thus providing significant computation time reduction over one-versus-all approaches. Furthermore, the leveraging weights enable effective data selection, thus reducing the cost of k-NN search at classification time. On the other hand, we propose a kernel generalization of our approach to take into account real-valued similarities between data in the feature space, thus enabling more accurate estimation of the local class density. We tested MLNN on three datasets of natural images. Results show that MLNN significantly outperforms classic k-NN and weighted k-NN voting. Furthermore, using an adaptive Gaussian kernel provides significant performance improvement. Finally, the best results are obtained when using MLNN with an appropriate learned metric distance.
Proceedings of the 10th Asian Conference on Computer Vision, ACCV 2010, November 8-12, 2010, Queenstown, New Zealand
Queenstown, New Zealand

hal-00664606
https://hal.inria.fr/hal-00664606
https://hal.inria.fr/hal-00664606/document
https://hal.inria.fr/hal-00664606/file/mlnn_accvfinal.pdf

Éditeur(s) : HAL CCSD IRD
Résumé : International audience
The number of satellites and remote sensing sensors devoted to earth observation becomes increasingly high, providing more and more data and especially images. In the same time the access to such a data and to the tools to process them has been considerably improved. In the presence of such data flow - and regarding the necessity to follow up and predict environmental and societal changes in highly dynamic socio-environmental contexts - we need automatic image interpretation methods. This could be accomplished by exploring some strengths of artificial intelligence. Our main idea consists in inducing classification rules that explicitly take into account structural knowledge, using Aleph, an Inductive Logic Programming (ILP) system. We applied our proposed methodology to three land cover/use maps of the French Guiana littoral. One hundred and forty six classification rules were induced for the 39 land-cover classes of the maps. These rules are expressed in first order logic language which make them intelligible and interpretable by non-experts. A ten-fold cross validation gave average values for classification accuracy, specificity and sensibility equal to, respectively, 98.82 %, 99.65% and 70%. The proposed methodology could be valuably exploited to automatically classify new objects and/or help operators using object-based classification procedures.
Symposium of the Latin American Society for Remote Sensing and Spatial Information Systems (SELPER)
Cayenne, French Guiana

ird-00797337
http://hal.ird.fr/ird-00797337
http://hal.ird.fr/ird-00797337/document
http://hal.ird.fr/ird-00797337/file/Bayoudh_et_al_2012_FP_SELPER-086.pdf
IRD : fdi:010058343

Éditeur(s) : HAL CCSD
Résumé : A standard approach for large scale image classification involves high dimensional features and Stochastic Gradient Descent algorithm (SGD) for the minimization of classical Hinge Loss in the primal space. Although complexity of Stochastic Gradient Descent is linear with the number of samples these method suffers from slow convergence. In order to cope with this issue, we propose here a Stochastic Low-Rank Newton Descent SLND for minimization of any calibrated loss in the primal space. SLND approximates the inverse Hessian by the best low-rank approximation according to squared Frobenius norm. We provide core optimization for fast convergence. Theoretically speaking, we show explicit convergence rates of the algorithm using these calibrated losses, which in addition provide working sets of parameters for experiments. Experiments are provided on the SUN, Caltech256 and ImageNet databases, with simple, uniform and efficient ways to tune remaining SLND parameters. On each of these databases, SLND challenges the accuracy of SGD with a speed of convergence faster by order of magnitude.
https://hal.archives-ouvertes.fr/hal-00825414

hal-00825414
https://hal.archives-ouvertes.fr/hal-00825414
https://hal.archives-ouvertes.fr/hal-00825414/document
https://hal.archives-ouvertes.fr/hal-00825414/file/TechnicalReport.pdf

Éditeur(s) : HAL CCSD Elsevier
Résumé : International audience
The number of satellites and sensors devoted to Earth observation has become increasingly elevated, delivering extensive data, especially images. At the same time, the access to such data and the tools needed to process them has considerably improved. In the presence of such data flow, we need automatic image interpretation methods, especially when it comes to the monitoring and prediction of environmental and societal changes in highly dynamic socio-environmental contexts. This could be accomplished via artificial intelligence. The concept described here relies on the induction of classification rules that explicitly take into account structural knowledge, using Aleph, an Inductive Logic Programming (ILP) system, combined with a multi-class classification procedure. This methodology was used to monitor changes in land cover/use of the French Guiana coastline. One hundred and fifty-eight classification rules were induced from 3 diachronic land cover/use maps including 38 classes. These rules were expressed in first order logic language, which makes them easily understandable by non-experts. A 10-fold cross-validation gave significant average values of 84.62%, 99.57% and 77.22% for classification accuracy, specificity and sensitivity, respectively. Our methodology could be beneficial to automatically classify new objects and to facilitate object-based classification procedures.
ISSN: 0098-3004

hal-01369822
https://hal.archives-ouvertes.fr/hal-01369822
https://hal.archives-ouvertes.fr/hal-01369822/document
https://hal.archives-ouvertes.fr/hal-01369822/file/CAGEO-D-13-00538R1_article.pdf
DOI : 10.1016/j.cageo.2014.08.013

Éditeur(s) : HAL CCSD Springer Verlag
Résumé : A preliminary version appeared in the 18th ACM-SIAM Symposium on Discrete Algorithms, pp. 746- 755, 2007
International audience
The Voronoi diagram of a finite set of objects is a fundamental geometric structure that subdivides the embedding space into regions, each region consisting of the points that are closer to a given object than to the others. We may define various variants of Voronoi diagrams depending on the class of objects, the distance function and the embedding space. In this paper, we investigate a framework for defining and building Voronoi diagrams for a broad class of distance functions called Bregman divergences. Bregman divergences include not only the traditional (squared) Euclidean distance but also various divergence measures based on entropic functions. Accordingly, Bregman Voronoi diagrams allow one to define information-theoretic Voronoi diagrams in sta- tistical parametric spaces based on the relative entropy of distributions. We define several types of Bregman diagrams, establish correspondences between those diagrams (using the Legendre transformation), and show how to compute them efficiently. We also introduce extensions of these diagrams, e.g. k-order and k-bag Bregman Voronoi diagrams, and introduce Bregman triangulations of a set of points and their connection with Bregman Voronoi diagrams. We show that these triangulations capture many of the properties of the celebrated Delaunay triangulation.
ISSN: 0179-5376

hal-00488441
https://hal.archives-ouvertes.fr/hal-00488441
https://hal.archives-ouvertes.fr/hal-00488441/document
https://hal.archives-ouvertes.fr/hal-00488441/file/BVD-DCG.pdf

Éditeur(s) : HAL CCSD
Résumé : International audience
In this paper, we investigate image segmentation by region merging. Given any similarity measure between regions , satisfying some weak constraints, we give a general predicate for answering if two regions are to be merged or not during the segmentation process. Our predicate is generic and has six properties. The first one is its inde-pendance with respect to the similarity measure, that leads to a user-independant and adaptative predicate. Second, it is non-parametric, and does not rely on any assumption concerning the image. Third, due to its weak constraints, knowledge may be included in the predicate to fit better to the user's behaviour. Fourth, provided the similarity is well-chosen by the user, we are able to upperbound one type of error made during the image segmentation. Fifth, it does not rely on a particular segmentation algorithm and can be used with almost all region-merging algorithms in various application domains. Sixth, it is calculated quickly, and can lead with appropriated algorithms to very efficient segmentation.
10th IEEE International Conference on Tools with Artificial Intelligence
Taipei, Taiwan

lirmm-01168371
https://hal-lirmm.ccsd.cnrs.fr/lirmm-01168371
https://hal-lirmm.ccsd.cnrs.fr/lirmm-01168371/document
https://hal-lirmm.ccsd.cnrs.fr/lirmm-01168371/file/FioNocICTAI98_Image%20Segmentation%20Using.pdf
DOI : 10.1109/TAI.1998.744885

Éditeur(s) : HAL CCSD
Résumé : International audience
Le nombre de satellites et de capteurs pour la télédétection dédiés à l'observation de la Terre ne cesse d'augmenter, permettant ainsi d'avoir une masse de données importante en particulier en matière d'images. Parallèlement, un effort permanent vise, d'une part, à améliorer l'accès à ces données et, d'autre part, à développer d'avantages d'outils pour les manipuler. De tels efforts sont particulièrement utiles dans des contextes socio-environnementaux très dynamiques spatialement et temporellement, pour lesquels il est nécessaire de suivre et de prédire les événements environnementaux et sociétaux. En revanche, en présence d'un tel flux de données, nous avons besoin de méthodes automatiques d'interprétation d'images. Une solution envisageable pour répondre à ce besoin est de bénéficier des atouts de l'intelligence artificielle pour l'obtention de cartes d'occupation du sol issues d'une classification des régions des images. Afin de contribuer à l'automatisation de la classification, nous proposons une méthode d'induction de règles interprétables par des non-experts et mettant en évidence, explicitement, des connaissances structurelles. Cette méthode s'appuie sur la programmation logique inductive (PLI) et en particulier sur le système inductif ”Aleph”. L'application de la méthode de classification Multi-class Rule Set Intersection (MRSI) permet ensuite de classifier tout nouvel objet au regard des ses caractéristiques intrinsèques et de celles des objets environnants. Nous avons appliqué notre méthodologie à l'étude de la dynamique du littoral de la Guyane Française. Suite à ce travail, nous avons induit 136 règles de classification pour 38 classes d'occupation du sol. Ces règles sont intelligibles et simples à interpréter de par l'utilisation de la logique du premier ordre. Les performances du système ont été évaluées par la validation croisée. En moyenne, la précision, la spécificité et la sensibilité sont, respectivement, égales à 84,62%, 99,57% et 77,22%. Ces résultats quantitatifs montrent une bonne performance de la méthodologie pour la mise à jour automatique de cartes d'occupation du sol et/ou l'assistance aux opérateurs utilisant l'analyse d'image orientée-objet.
11èmes Rencontres des Jeunes Chercheurs en Intelligence Artificielle
Lille, France

hal-01376400
https://hal.archives-ouvertes.fr/hal-01376400
https://hal.archives-ouvertes.fr/hal-01376400/document
https://hal.archives-ouvertes.fr/hal-01376400/file/Bayoudh_RJCIA2013.pdf

Éditeur(s) : HAL CCSD IEEE
Résumé : International audience
This paper proposes a novel automated approach for the categorization of cells in fluorescence microscopy images. Our supervised classification method aims at recognizing patterns of unlabeled cells based on an annotated dataset. First, the cell images need to be indexed by encoding them in a feature space. For this purpose, we propose tailored bio-inspired features relying on the distribution of contrast information. Then, a supervised learning algorithm is proposed for classifying the cells. We carried out experiments on cellular images related to the diagnosis of autoimmune diseases, testing our classification method on the HEp-2 Cells dataset of Foggia et al (CBMS 2010). Results show classification precision larger than 96% on average, thus confirming promising application of our approach to the challenging application of cellular image classification for computer-aided diagnosis.
ICPR 2012
Tsukuba, Japan

hal-00958856
https://hal.archives-ouvertes.fr/hal-00958856
https://hal.archives-ouvertes.fr/hal-00958856/document
https://hal.archives-ouvertes.fr/hal-00958856/file/icpr_2012.pdf

Éditeur(s) : HAL CCSD Institute of Electrical and Electronics Engineers
Résumé : International audience
Tailoring nearest neighbors algorithms to boosting is an important problem. Recent papers study an approach, UNN, which provably minimizes particular convex surrogates under weak assumptions. However, numerical issues make it necessary to experimentally tweak parts of the UNN algorithm, at the possible expense of the algorithm's convergence and performance. In this paper, we propose a lightweight alternative algorithm optimizing proper scoring rules from a very broad set, and establish formal convergence rates under the boosting framework that surprisingly compete with those known for UNN. It is an adaptive Newton-Raphson algorithm, which belongs to the same lineage as the popular Gentle Adaboost. To the best of our knowledge, no such boosting-compliant convergence rates were previously known for these algorithms. We provide experiments on a dozen domains, including the challenging Caltech and SUN computer vision databases. They display that GNNB significantly outperforms UNN, both in terms of convergence rate and quality of the solution obtained, and GNNB provides a simple and efficient contender to techniques that can be used on very large domains, like stochastic gradient descent -- for which little is known to date. Experiments include a divide-and-conquer improvement of GNNB which exploits the link with proper scoring rules optimization.
ISSN: 0162-8828

hal-00958809
https://hal.archives-ouvertes.fr/hal-00958809
https://hal.archives-ouvertes.fr/hal-00958809/document
https://hal.archives-ouvertes.fr/hal-00958809/file/PAMI.pdf
DOI : 10.1109/TPAMI.2014.2307877

Éditeur(s) : HAL CCSD
Résumé : International audience
It is an admitted fact that mainstream boosting algorithms like AdaBoost do not perform well to estimate class conditional probabilities. In this paper, we analyze, in the light of this problem, a recent algorithm, unn, which leverages nearest neighbors while minimizing a convex loss. Our contribution is threefold. First, we show that there exists a subclass of surrogate losses, elsewhere called balanced, whose minimization brings simple and statistically efficient estimators for Bayes posteriors. Second, we show explicit convergence rates towards these estimators for unn, for any such surrogate loss, under a Weak Learning Assumption which parallels that of classical boosting results. Third and last, we provide experiments and comparisons on synthetic and real datasets, including the challenging SUN computer vision database. Results clearly display that boosting nearest neighbors may provide highly accurate estimators, sometimes more than a hundred times more accurate than those of other contenders like support vector machines.
European Conference on Machine Learning and Principles and Practice of Knowledge Discovery in Databases (ECML PKDD)
Bristol, United Kingdom

hal-00702771
https://hal.archives-ouvertes.fr/hal-00702771
https://hal.archives-ouvertes.fr/hal-00702771/document
https://hal.archives-ouvertes.fr/hal-00702771/file/ecml12-dnbnb-sub.pdf

Éditeur(s) : HAL CCSD
Résumé : The Voronoi diagram of a finite set of objects is a fundamental geometric structure that subdivides the embedding space into regions, each region consisting of the points that are closer to a given object than to the others. We may define many variants of Voronoi diagrams depending on the class of objects, the distance functions and the embedding space. In this paper, we investigate a framework for defining and building Voronoi diagrams for a broad class of distance functions called Bregman divergences. Bregman divergences include not only the traditional (squared) Euclidean distance but also various divergence measures based on entropic functions. Accordingly, Bregman Voronoi diagrams allow to define information-theoretic Voronoi diagrams in statistical parametric spaces based on the relative entropy of distributions. We define several types of Bregman diagrams, establish correspondences between those diagrams (using the Legendre transformation), and show how to compute them efficiently. We also introduce extensions of these diagrams, e.g. k-order and k-bag Bregman Voronoi diagrams, and introduce Bregman triangulations of a set of points and their connexion with Bregman Voronoi diagrams. We show that these triangulations capture many of the properties of the celebrated Delaunay triangulation. Finally, we give some applications of Bregman Voronoi diagrams which are of interest in the context of computational geometry and machine learning.
https://hal.inria.fr/inria-00137865

Report N° : RR-6154
inria-00137865
https://hal.inria.fr/inria-00137865
https://hal.inria.fr/inria-00137865v2/document
https://hal.inria.fr/inria-00137865/file/bregman.pdf

Éditeur(s) : HAL CCSD Springer
Résumé : Third International Workshop, MLMI 2012, Held in Conjunction with MICCAI 2012
International audience
Universal Nearest Neighbours (unn) is a classifier recently proposed, which can also effectively estimates the posterior probability of each classification act. This algorithm, intrinsically binary, requires the use of a decomposition method to cope with multiclass problems, thus reducing their complexity in less complex binary subtasks. Then, a reconstruction rule provides the final classification. In this paper we show that the application of unn algorithm in conjunction with a reconstruction rule based on the posterior probabilities provides a classification scheme robust among different biomedical image datasets. To this aim, we compare unn performance with those achieved by Support Vector Machine with two different kernels and by a k Nearest Neighbours classifier, and applying two different reconstruction rules for each of the aforementioned classification paradigms. The results on one private and five public biomedical datasets show satisfactory performance.
Machine Learning in Medical Imaging
Nice, France

hal-00958909
https://hal.archives-ouvertes.fr/hal-00958909
https://hal.archives-ouvertes.fr/hal-00958909/document
https://hal.archives-ouvertes.fr/hal-00958909/file/Miccai.pdf
DOI : 10.1007/978-3-642-35428-1_15

Éditeur(s) : HAL CCSD Springer Verlag
Résumé : under revision for IJCV
International audience
The k-nearest neighbors (k-NN) classification rule has proven extremely successful in countless many computer vision applications. For example, image categorization often relies on uniform voting among the nearest prototypes in the space of descriptors. In spite of its good properties, the classic k-NN rule suffers from high variance when dealing with sparse prototype datasets in high dimensions. A few techniques have been proposed to improve k-NN classification, which rely on either deforming the nearest neighborhood relationship or modifying the input space. In this paper, we propose a novel boosting algorithm, called UNN (Universal Nearest Neighbors), which induces leveraged k-NN, thus generalizing the classic k-NN rule. We redefine the voting rule as a strong classifier that linearly combines predictions from the k closest prototypes. Weak classifiers are learned by UNN so as to minimize a surrogate risk. A major feature of UNN is the ability to learn which prototypes are the most relevant for a given class, thus allowing one for effective data reduction. Experimental results on the synthetic two-class dataset of Ripley show that such a filtering strategy is able to reject "noisy" prototypes. We carried out image categorization experiments on a database containing eight classes of natural scenes. We show that our method outperforms significantly the classic k-NN classification, while enabling significant reduction of the computational cost by means of data filtering.
ISSN: 0920-5691

hal-00481712
https://hal.archives-ouvertes.fr/hal-00481712
DOI : 10.1007/s11263-012-0539-2
ARXIV : 1001.1221

Éditeur(s) : HAL CCSD IEEE
Résumé : Object classification is a challenging task in computer vision. Many approaches have been proposed to extract meaningful descriptors from images and classifying them in a supervised learning framework. In this paper, we revisit the classic k-nearest neighbors (k-NN) classification rule, which has shown to be very effective when dealing with local image descriptors. However, k-NN still features some major drawbacks, mainly due to the uniform voting among the nearest prototypes in the feature space. In this paper, we propose a generalization of the classic k-NN rule in a supervised learning (boosting) framework. Namely, we redefine the voting rule as a strong classifier that linearly combines predictions from the k closest prototypes. To induce this classifier, we propose a novel learning algorithm, MLNN (Multiclass Leveraged Nearest Neighbors), which gives a simple procedure for performing prototype selection very efficiently. We tested our method on 12 categories of objects, and observed significant improvement over classic k-NN in terms of classification performances.
International Workshop on Image Analysis for Multimedia Interactive Services
Desenzano del Garda, Italy

hal-00481725
https://hal.archives-ouvertes.fr/hal-00481725
https://hal.archives-ouvertes.fr/hal-00481725/document
https://hal.archives-ouvertes.fr/hal-00481725/file/pbnn_wiamis_10.pdf

Éditeur(s) : HAL CCSD Elsevier
Résumé : International audience
A data stream is a potentially uninterrupted flow of data. Mining this flow makes it necessary to cope with uncertainty, as only a part of the stream can be stored. In this paper, we evaluate a statistical technique which biases the estimation of the support of patterns, so as to maximize either the precision or the recall, as chosen by the user, and limit the degradation of the other criterion. Theoretical results show that the technique is not far from the optimum, from the statistical standpoint. Experiments performed tend to demonstrate its potential, as it remains robust even under significant distribution drifts.
ISSN: 0031-3203

lirmm-00365474
https://hal-lirmm.ccsd.cnrs.fr/lirmm-00365474

Éditeur(s) : HAL CCSD Elsevier
Résumé : International audience
Recently, a large amount of work has been devoted to the study of spectral clustering — a powerful unsupervised classification method. This paper brings contributions to both its foundations, and its applications to text classification. Departing from the mainstream, concerned with hard membership, we study the extension of spectral clustering to soft membership (probabilistic, EM style) assignments. One of its key features is to avoid the complexity gap of hard membership. We apply this theory to a challenging problem, text clustering for languages having permeable borders, via a novel construction of Markov chains from corpora. Experiments with a readily available code clearly display the potential of the method, which brings a visually appealing soft distinction of languages that may define altogether a whole corpus.
ISSN: 0031-3203

hal-00329991
https://hal.archives-ouvertes.fr/hal-00329991
DOI : 10.1016/j.patcog.2008.06.024