Search

IFSA-EUSFLAT'2015: 16th World Congress of the International Fuzzy Systems Association and 9th Conference of the European Society for Fuzzy Logic and Technlogy, July 2015, Gijón, Spain, We propose a new point of view in the long-standing problem where several voters have expressed a linear order relation (or ranking) over a set of candidates. For a ranking a > b > c to represent a group’s opinion, it would be logical that the strength with which a > c is supported should not be less than the strength with which either a > b or b > c is supported. This intuitive property can be considered a monotonicity constraint, and has been addressed before. We extend previous approaches in the following way: as the voters are expressing linear orders, we can take the number of candidates between two candidates to be a measure of the degree to which one candidate is preferred to the other. In this way, intensity of support is both counted as the number of voters who indicate a > c is true, as well as the distance between a and c in these voters’ rankings. The resulting distributions serve as input for a natural ranking rule that is based on stochastic monotonicity and stochastic dominance. Adapting the previous methodology turns out to be non-trivial once we add some natural feasibility constraints, This work has been partially supported by Campus of International Excellence of University of Oviedo

Transfer learning provides an approach to solve target tasks more quickly and effectively by using previouslyacquired knowledge learned from source tasks. Most of transfer learning approaches extract knowledge of source domain in the given feature space. The issue is that single perspective can‟t mine the relationship of source domain and target domain fully. To deal with this issue, this paper develops a method using Stacked Denoising Autoencoder (SDA) to extract new feature spaces for source domain and target domain, and define two fuzzy sets to analyse the variation of prediction accuracy of target task in new feature spaces.

In this work we construct linear orders between pairs of intervals by using aggregation functions. We apply these orders in a decision-making problem where the experts provide their opinions by means of interval-valued fuzzy sets.

The notion of extensionality of a fuzzy relation w.r.t. a fuzzy equivalence was first introduced by Hohle and Blanchard. Belohlavek introduced a similar definition of compatibility of a fuzzy relation w.r.t. a fuzzy equality. In [14] we generalized this notion to left compatibility, right compatibility and compatibility of arbitrary fuzzy relations and we characterized them in terms of left and right traces introduced by Fodor. In this note, we will again investigate these notions, but this time we focus on the compatibility of strict orders with fuzzy tolerance and fuzzy equivalence relations.

Digital images are generally created as discrete measurements of light, as performed by dedicated sensors. Consequently, each pixel contains a discrete approximation of the light inciding in a sensor element. The nature of this measurement implies certain uncertainty due to discretization matters. In this work we propose to model such uncertainty using intervals, further leading to the generation of so-called interval-valued images. Then, we study the partial differentiation of such images, putting a spotlight on antisymmetric convolution operators for such task. Finally, we illustrate the utility of the interval-valued images by studying the behaviour of an extended version of the well-known Canny edges detection method.

In the present paper, we connect some old reflections about the relationships existing between the theory of fuzzy sets and cybernetics with modern, contemporary analyses of the crucial (better: unavoidable) role that fuzziness plays in the attempts at scientifically describing aspects of information sciences. The connection, which has a basic conceptual origin, has been triggered also by the recent 50th anniversary of Norbert Wiener’s death which has been instrumental in looking again at some crucial aspects of the birth of information sciences in the midst of the last century. Fuzzy sets are an essential part of this revolution and share all the innovations as well as the difficulties of this towering scientific enterprise which has changed the vision of what a scientific approach must be when dealing with something like information so different from the old matter and energy.t These considerations are helpful in looking in an enlarged way at how to treat and consider the notion of cognition.

In the present paper, we connect some old reflections about the relationships existing between the theory of fuzzy sets and cybernetics with modern, contemporary analyses of the crucial (better: unavoidable) role that fuzziness plays in the attempts at scientifically describing aspects of information sciences. The connection, which has a basic conceptual origin, has been triggered also by the recent 50th anniversary of Norbert Wiener’ death which has been instrumental in looking again at some crucial aspects of the birth of information sciences in the midst of last Century. Fuzzy sets are an essential part of this revolution and share all the innovations as well as the difficulties of this towering scientific enterprise which has changed the vision of what a scientific approach must be when dealing with something like information so different from the old matter and energy. These considerations are helpful in looking in an enlarged way at how treat and consider the notion of cognition.

This paper presents a comparative analysis of the performance of fuzzy approaches on the task of predicting customer review ratings using a computational intelligence framework based on a genetic algorithm for data dimensionality reduction. The performance of the Fuzzy C-Means (FCM), a neurofuzzy approach combining FCM and the Adaptive Neuro Fuzzy Inference System (ANFIS), and the Simplified Fuzzy ARTMAP (SFAM) was compared on six datasets containing customer reviews. The results revealed that all computational intelligence predictors were suitable for the rating prediction problem, and that the genetic algorithm is effective in reducing the number of dimensions without affecting the prediction performance of each computational intelligence predictor.

En este trabajo se analiza una cuestión clave respecto de la relación que debe existir entre operadores de una misma familia de operadores de agregación (FAO) {An}, a fin de comprender que ellos deben definir adecuadamente un todo consistente. Se extienden algunas de las ideas de estabilidad de una FAO con un enfoque más general,definiéndose formalmente las nociones de i−L y j−R estabilidad estricta para familias de operadores de agregación, e introduciendo la noción de estabilidad estricta de orden k. Finalmente,se muestra una aplicación de las condiciones de estabilidad estricta al problema de pérdida de datos en un proceso de agregación de información, utilizándose las familias de la media ponderada y de la media ponderada cuasi-aritmética.

En este artículo se presentan los resultados de una experiencia realizada con dos grupos de alumnos,uno de Ciencias Sociales y otro de Ciencias Matemáticas, con la finalidad de observar las posibles disimilitudes en la percepción y uso de diferentes lógicas. El resultado de esta experiencia inicial mostrará, sobre todo, cómo el científico acepta más fácilmente que el humanista modelos que su propio sentido común rechaza, lo cual es un aspecto que se debe tomar en cuenta en estudios formales posteriores.

Extraction of rules for classification and decision tasks from databases is an issue of growing importance as automated processes based on data are being required in these fields. Interpretability of rules is improved by defining classes for independent variables. Moreover, though more complex, a more realistic and flexible framework is attained when fuzzy classes are considered. In this paper, an inductive approach is taken in order to develop a general methodology for building fuzzy rules from databases. Three types of rules are built in order to be able of dealing with both categorical and numerical data.

Although consistency in information and actions is a major argument in any decision making problem, most available tools are crisp, while observation strongly suggests to consider different degrees of consistency, i.e.,consistency is fuzzy in nature. In this paper we propose to put together some of the works in the field, focusing on information consistency rather than on action consistency, showing in this way key classification features.

The main contribution of this paper is to obtain a polynomial algorithm that allows us to classify step by step a set of items that are related by means of a fuzzy graph extending the algorithm published in [8] to any class of graph.

In this contribution some applications of Fuzzy Set Theory to Information Retrieval are described, as well as the more recent outcomes of this research field. Fuzzy Set Theory is applied to Information Retrieval to the main aim to define flexible systems, i.e. systems that can represent and manage the vagueness and subjectivity which characterizes the process of information representation and retrieval.