Gustavo J. Meschino

Reconfiguration of electrical networks by an Ant Colony Optimization algorithm

Electricity distribution companies constantly require improvements in service, and an appropriate reduction in costs of the system. Finding the optimal configuration of the distribution network reduces power losses, which impacts directly on costs, also leading to significant energy savings. Over recent decades, the problem of network reconfiguration for loss minimization and for other purposes has been widely studied. The algorithms improve the previously obtained minimum power loss and also reduce computational times. This paper proposes the study and application of an algorithm based on Ant Colony Optimization, method covered in the paradigm of Swarm Intelligence, sub-discipline of Computational Intelligence. We show an easy network codification and we performed a detailed study of the ranges of values for the configuration parameters of the algorithm. We achieved the optimal result for testing networks and we found appropriate configurations for more complex networks, taking low computational times.

Automatic design of interpretable fuzzy predicate systems for clustering using self-organizing maps

Abstract In the area of pattern recognition, clustering algorithms are a family of unsupervised classifiers designed with the aim to discover unrevealed structures in the data. While this is a never ending research topic, many methods have been developed with good theoretical and practical properties. One of such methods is based on self organizing maps (SOM), which have been successfully used for data clustering, using a two levels clustering approach. Newer on the field, clustering systems based on fuzzy logic improve the performance of traditional approaches. In this paper we combine both approaches. Most of the previous works on fuzzy clustering are based on fuzzy inference systems, but we propose the design of a new clustering system in which we use predicate fuzzy logic to perform the clustering task, being automatically designed based on data. Given a datum, degrees of truth of fuzzy predicates associated with each cluster are computed using continuous membership functions defined over data features. The predicate with the maximum degree of truth determines the cluster to be assigned. Knowledge is discovered from data, obtained using the SOM generalization aptitude and taking advantage of the well-known SOM abilities to discover natural data grouping when compared with direct clustering. In addition, the proposed approach adds linguistic interpretability when membership functions are analyzed by a field expert. We also present how this approach can be used to deal with partitioned data. Results show that clustering accuracy obtained is high and it outperforms other methods in the majority of datasets tested.

Optimización de Redes Eléctricas Mediante la Aplicación de Algoritmos Genéticos

This work proposes the optimization of the topological configuration of secondary electrical distribution networks to minimize technical losses produced during their exploitation due to Joule effect, using for this Genetic Algorithms. Through the application of the method to two systems, it was found that the optimization technique is able to find the optimum solution among all the possible combinations that switch operations offer. Its flexibility to adapt to the restrictions of radiality and voltage level in less time than that of exhaustive search is also proved. During the development of the application the genetic operators were validated, determining those that gave the best performance in the search of the solution. The results show the feasibility and viability of the application for the optimum configuration of electrical distribution systems.

Probability mapping images in dynamic speckle classification

We propose the use of a learning procedure to identify regions of similar dynamics in speckle image sequences that includes more than one descriptor. This procedure is based on the application of a naïve Bayes statistical classifier comprising the use of several descriptors. The class frontiers can be depicted so that the proportion of identified regions may be measured. To demonstrate the results, assembly of an RGB image, where each plane (R, G, and B) is associated with a particular region (class), was labeled according to its biospeckle dynamics. A high brightness in one color means a high probability of the pixel belonging to the corresponding class, and vice versa.

Dynamic laser speckle: decision models with computational intelligence techniques

This paper proposes the design of decision models with Computational Intelligence techniques using image sequences of dynamic laser speckle. These models aim to characterize the dynamic of the process evaluated through Temporal History Speckle Patterns (THSP) using a set of available descriptors. The models use those sets selected to improve its effectiveness, depending on the specific application. The techniques of computational intelligence field include using Artificial Neural Networks, Fuzzy Granular Computation, Evolutionary Computation elements such as Genetic Algorithms, among others. The results obtained in experiments such as the evaluation of bacterial chemotaxis, and the estimation of the drying time of coatings are encouraging and significantly improve those obtained using a single descriptor.

Biospeckle image stack process based on artificial neural networks

This paper proposes the identification of regions of interest in biospeckle patterns using unsupervised neural networks of the type Self-Organizing Maps. Segmented images are obtained from the acquisition and processing of laser speckle sequences. The dynamic speckle is a phenomenon that occurs when a beam of coherent light illuminates a sample in which there is some type of activity, not visible, which results in a variable pattern over time. In this particular case the method is applied to the evaluation of bacterial chemotaxis. Image stacks provided by a set of experiments are processed to extract features of the intensity dynamics. A Self-Organizing Map is trained and its cells are colored according to a criterion of similarity. During the recall stage the features of patterns belonging to a new biospeckle sample impact on the map, generating a new image using the color of the map cells impacted by the sample patterns. It is considered that this method has shown better performance to identify regions of interest than those that use a single descriptor. To test the method a chemotaxis assay experiment was performed, where regions were differentiated according to the bacterial motility within the sample.

Type-2 Fuzzy Logic in Decision Support Systems

Decision Support Systems have been widely used in expert knowledge modeling. One of the known implementation approaches is through definition of Fuzzy Sets and Fuzzy Predicates, whose evaluation determines the system’s output. Despite Type-1 Fuzzy Sets have been widely used in this type of implementation, there are uncertainty sources that cannot be adequately modeled when using expert knowledge minimizing their effect on system’s output, especially when it comes from several experts opinions. Type-2 Fuzzy Sets deal with fuzzy membership degrees, which can represent adequately the typical uncertainties of these systems. In this chapter, we generalize the operators of Fuzzy Logic in order to evaluate Fuzzy Predicates with Type-2 Fuzzy Sets and we define measures to assess the degree of truth of these predicates to define the theoretical background of the Decision Support Systems using this methodology. We present an example application of decision-making and a brief discussion of the results.

Using SOM as a Tool for Automated Design of Clustering Systems Based on Fuzzy Predicates

Clustering task is a never-ending research topic. New methods are permanently proposed. In particular, Fuzzy Logic and Self-organizing Maps and their mutual cooperation have demonstrated to be interesting paradigms. We propose a general approach to obtain membership functions for a ranked clustering system based on fuzzy predicates logical operations, considering Gaussian-shaped curves. We find membership functions parameters from trained Self-organizing Maps, which generalize the statistical characteristics of data. The system is self-configured and it has the advantages of other fuzzy approaches. Clustering quality is assessed by labeled data, which allow computing accuracy. The proposal must be tested with more real datasets, though the preliminary results obtained in well-known datasets suggest that it is a promising clustering scheme.

Linguistic Interpretation of Mathematical Morphology

Mathematical Morphology is a theory based on geometry, algebra, topology and set theory, with strong application to digital image processing. This theory is characterized by two basic operators: dilation and erosion. In this work we redefine these operators based on compensatory fuzzy logic using a linguistic definition, compatible with previous definitions of Fuzzy Mathematical Morphology. A comparison to previous definitions is presented, assessing robustness against noise.

Discovering knowledge from data clustering using automatically-defined interval type-2 fuzzy predicates

It is proposed a new clustering method based on interval type-2 fuzzy predicates. Fuzzy predicates are automatically generated from data describing clusters. Interval type-2 membership functions model variability and vagueness in clusters. Linguistic descriptions and knowledge are extracted from predicates. The method can be applied to data analysis applications. In data clustering fuzzy predicates act as cluster descriptors providing linguistically expressed knowledge which indicates how features are related to each cluster. Fuzzy predicates directly and automatically obtained from data enable discovering knowledge inside clusters, even when there is no prior-information about the clustering problem. In this work a new method for automatic discovering of interval type-2 fuzzy predicates in data clustering is proposed, called Type-2 Data-based Fuzzy Predicate Clustering (T2-DFPC). In a first stage, a data analysis is performed by making a random partition of the original data and running a clustering scheme that automatically determines the suitable number of clusters. From this stage, interval type-2 fuzzy predicates are discovered. Results obtained on very different clustering datasets show that the T2-DFPC method was consistently one of the best in terms of accuracy. The method preserves all known advantages of the interval type-2 FL to deal with problems with vagueness, quantifying the degree of truth of the fuzzy predicates and modelling the variability of the data inside the clusters. The proposed method is a fast, useful, general, and unsupervised approach for interpretable data clustering, being the knowledge-extracting capabilities one of the main contributions. Linguistic expressions can be easily adapted to match the terminology used in the field the data are related to. The predicates are able to generalize the knowledge for new cases (new data), as an intelligent system. This new approach might be surprisingly useful in contexts where, besides the clustering partition, summary information from data is of interest.