1,720,956 research outputs found
Estudio e implementación en MOA de una nueva propuesta del método Bagging para combinar clasificadores base
No full textEn la actualidad, el volumen de los datos generados por sensores, Internet, dispositivos de localización, telefonía y muchos otros, esta en constante aumento. El tamaño de estos datos es potencialmente infinito debido a su constante generación, por lo que es necesario procesarlos de forma eficiente utilizando un límite escaso de memoria en un límite reducido de tiempo. Los Sistemas Multiclasificadores proporcionan un mecanismo eficaz de combinar los resultados de un conjunto de clasificadores obteniendo un modelo más complejo que un clasificador simple logrando mejor precisión en la predicción. Existen varios tipos de multiclasificadores en este trabajo se profundiza en el estudio de los multiclasificadores incrementales Bagging y Boosting, el método Bagging combina los resultados de clasificadores entrenados con conjuntos de entrenamientos generados por la técnica de bootstraping, los clasificadores obtenidos mediante esta técnica son independientes entre sí. Bagging obtiene los resultados de la clasificación a través de una votación siguiendo el consenso de prulabilidad. En estudios realizados en el año 1997 Lam y Suen [30] estudiaron el voto mayoritario como método de combinación de un número par o impar de clasificadores que pueden votar por una de las k clases o abstenerse. Con el estudio del voto mayoritario obtienen algunos resultados interesantes, llegan a la conclusión que el voto mayoritario no es una buena técnica de combinación cuando nos encontramos en presencia de clasificadores individuales. Análisis Masivo Online(MOA) es un entorno de software para la implementación de algoritmos permitiéndonos llevar a cabo experimentos para el aprendizaje en línea con flujos de datos. MOA incluye una colección de métodos tradicionales y online, así como herramientas para la evaluaciófile. Entre estos métodos se encuentran versiones en linea del método Bagging, se ha hecho evidente la necesidad de estudiar y experimentar distintas formas de combinación para este método pues obtiene clasificadores independientes. Para dicho estudio se propone en este trabajo la realización de una nueva versión en linea del algoritmo Bagging para la herramienta MOA que no combine el resultado de los clasificadores mediante la votación mayoritaria, se proponen como métodos de combinación Naıve Bayes (NB), EWMA, Entropy Weighting (EW) y Dempster–Shafer (DS). Se logra incluir este nuevo método mediante la realización de un nuevo clasificador quedando implementado en el subpaquete moa.classifiers.meta, donde la clase principal que la representa se denomina AbstractClassifier y fue compilado con la versión 1.8 de la máquina virtual de Java. Para la evaluación de este nuevo método fue comparado empíricamente con varios algoritmos de ensamble de la familia bagging, utilizando Naive Bayes como clasificador base. Los experimentos incluyeron conjuntos de datos artificiales y reales. Todos los algoritmos fueron probados frente a los tipos de cambios comunes (abruptos y graduales). Los experimentos mostraron que el nuevo algoritmo es una buena opción para el aprendizaje a partir de flujos de datos con cambios
de concepto.At the present time, the volume of the data generated by sensors, Internet, localization devices, telephony and many other, this in constant increase. The size of these data is potentially infinite due to its constant generation, for what is necessary to process them in an efficient way using a scarce limit by heart in a reduced limit of time. The Systems Multiclasificadores provides an effective mechanism of combining the results of a group of classifiers obtaining a more complex model that a simple classifier achieving better precision in the prediction. Several multiclasificadores types exist in this work it is deepened in the study of the incremental multiclasificadores Bagging and Boosting, the method Bagging combines the results of classifiers trained with groups of trainings generated by the bootstraping technique, the classifiers obtained by means of this technique are independent to each other. Bagging obtains the results of the classification through a voting following the prulabilidad consent. In studies carried out in the year 1997 Lam and Suen [30] did they study the majority vote as method of combination of an even or odd number of classifiers that you/they can vote for one of those?? classes or to abstain. With the study of the majority vote they obtain some interesting results, they arrive to the conclución that the majority vote is not a good combination technique when we are in presence of individual classifiers. Massive analysis Online(MOA) it is a software environment for the implementation of algorithms allowing us to carry out experiments for the on-line learning with flows of data. MOA includes a collection of traditional methods and online, as well as tools for the evaluation. Among these methods they are versions in it lines of the method Bagging, it has become evident the necessity to study and to experience different combination forms for this since method obtains independent classifiers. For this study he/she intends in this work the realization of a new version in it lines of the algorithm Bagging for the tool MOA that doesn’t combine the result of the classifiers by means of the majority voting, they intend as combination methods Naive Bayes (NB), EWMA, EntropyWeighting (EW) and Dempster–Shafer (DS). it is possible to include this new method by means of the realization of a new classifier being implemented in the subpaquete moa.classifiers.meta, where the main class that represents it is denominated AbstractClassifier 10 and it was compiled with the version 1.8 of the virtual machine of Java. For the evaluation of this new method it was compared empirically with several algorithms of it assembles of the family bagging, using Naive Bayes like classifier bases. The experiments included groups of artificial and real data. All the algorithms were proven in front of the types of common changes (abrupt and gradual). The experiments showed that the new algorithm is a good option for the learning starting from flows of data with concept changes.Facultad de Matemática, Física y Computación. Departamento de Ciencias de la Computaciónnon-publishe
Detección de spam utilizando clasificadores incrementales
No full textEl correo electrónico, pese a su antigüedad, sigue siendo una forma popular de comunicación
por ser muy económico y fácil de usar. La mayoría de los correos electrónicos que son enviados
en la actualidad son correos spam los cuales suponen una pérdida de tiempo y dinero para usuarios
y empresas. Para contrarrestar el spam se han creado herramientas que intentan detectarlo de
forma automática. No obstante, no existe un método que permita la detección de spam con una
precisión absoluta dado que los spammers cambian la forma de los spam para confundir a los
filtros anti-spam. Sin embargo, en el área del aprendizaje automático existen algoritmos capaces
de reconocer estos patrones (conocido como cambio de concepto) y adaptarse al mismo. El presente
trabajo se propuso como objetivo detectar de manera eficiente los correos spam utilizando
clasificadores incrementales, de los cuales algunos son capaces de adaptarse al cambio de concepto,
para lo cual se diseñó un modelo para la detección personalizada de spam y se realizaron
experimentos para evaluar el desempeño de los algoritmos a utilizar en el modelo, obteniendo
resultados favorables que permitieron llegar a la conclusión de que este tipo de clasificadores
son eficaces para la detección de spam, con un elevado grado de precisión en las predicciones,
destacándose el clasificador FASE, el cual fue seleccionado para formar parte de la herramienta
final propuesta.Email, despite its age, still is a popular form of communication because it is very economic
and easy to use. Most of the emails that are sent today are spam emails, which represent a
waste of time and money for users and companies. To counteract spam, tools have been created
that try to detect it automatically. However, there is no method that allows the detection of
spam with absolute precision since spammers change the form of spam to confuse anti-spam
filters. However, in the machine learning area there are algorithms capable of recognizing these
patterns (also know as concept drift) and adapting to them. The objective of the present work is
to efficiently detect spam emails using online classifiers, some of which are capable of adapting
to concept drift. A model for the personalized spam detection was designed and experiments
were carried out to evaluate the performance of the algorithms to be used in the model, obtaining
favorable results that allowed to reach the conclusion that this type of classifier are effective for
spam detection, with a high percent of accuracy in the predictions, standing out from the rest
FASE classifier, which was selected to be part of the proposed final tool.Facultad de Matemática, Física y Computación. Departamento de Ciencias de la Computaciónnon-publishe
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
Dispelling the Myths Behind First-author Citation Counts
Get PDFWe conducted a full-scale evaluative citation analysis study of scholars in the XML research field to explore just how different from each other author rankings resulting from different citation counting methods actually are, and to demonstrate the capability of emerging data and tools on the Web in supporting more realistic citation counting methods. Our results contest some common arguments for the continued
use of first-author citation counts in the evaluation of scholars, such as high correlations between author rankings by first-author citation counts and other citation
counting methods, and high costs of using more realistic citation counting methods that are not well-supported by the ISI databases. It is argued that increasingly available digital full text research papers make it possible for citation analysis studies to go beyond what the ISI databases have directly supported and to employ more
sophisticated methods
koamabayili/VECTRON-author-checklist: VECTRON author checklist
No full textWe have done our best to complete the author checklist relating to the use of animals in the hut study. Note that the objective for the hut study was to evaluate the IRS treatment applications for residual efficacy against Anopheles mosquitoes, including the local An. coluzzii mosquito population. Cows were only used to attract mosquitoes into the huts and no tests were carried out directly on the cows. The author checklist is intended for use with studies where experiments are carried out on animals, which is why we have had such difficulty in completing this for the hut study, as many of the questions do not relate to how the cows were used
Author-wise bibliometric analysis based on entropy.
No full textAuthor-wise bibliometric analysis based on entropy.</p
- …
