J. Aroba

Identification of piecewise affine systems by means of fuzzy clustering and competitive learning

This paper presents an identification method for a class of dynamic system known as piecewise affine systems. Such systems are composed of a set of affine maps which relate inputs and outputs. These maps are defined in disjunctive regions in the regression space, itself composed of system inputs and outputs. The aim of the proposed method is to obtain a model of the system from a set of input-output data. This model comprises a set of submodels defined in different regions of the regression space. The proposed method is sequenced according to several stages which identify the set of submodels and the regions in which they are defined. These submodels are obtained by means of an algorithm inspired by competitive learning which rewards those that best fit the data in each region of the regression space. The method uses a process of fuzzy clustering in order to obtain a subset of representatives from the original data set, so reducing the amount of information to be processed while retaining the significant information from the original data and minimizing the effect of noise on the data.

Precipitation, pH and metal load in AMD river basins: an application of fuzzy clustering algorithms to the process characterization

In the present work, Acid Mine Drainage (AMD) processes in the Chorrito Stream, which flows into the Cobica River (Iberian Pyrite Belt, Southwest Spain) are characterized by means of clustering techniques based on fuzzy logic. Also, pH behavior in contrast to precipitation is clearly explained, proving that the influence of rainfall inputs on the acidity and, as a result, on the metal load of a riverbed undergoing AMD processes highly depends on the moment when it occurs. In general, the riverbed dynamic behavior is the response to the sum of instant stimuli produced by isolated rainfall, the seasonal memory depending on the moment of the target hydrological year and, finally, the own inertia of the river basin, as a result of an accumulation process caused by age-long mining activity.

Presence of As in the fluvial network due to AMD processes in the Riotinto mining area (SW Spain): A fuzzy logic qualitative model

The Tinto River crosses the mining area of Riotinto (Iberian Pyrite Belt, SW Spain), where it receives the highest contribution of contaminants (AMD). In this paper we apply a fuzzy computer tool, PreFuRGe, which allows qualitative interpretation of the data recorded in a database relating to the chemistry of water. Specifically, we aim to find information not likely to be detected by means of classical statistical techniques, and which can help in characterizing and interpreting the behavior of arsenic in a complex system. The conclusions present that the factors which most directly control the presence of total dissolved As are closely linked to the climate and are temperature and rainfall, and therefore pH. As (III) is also shown to be related to temperature and pH. In terms of temperature As (V) is found to operate in a way which is the opposite of As (III). In terms of pH the relationship is not as clear as for As (III). As for rain, the highest As (V) values are compatible with minimum or non-existent rainfall, while minimum values correspond to any value for rainfall, including very high.

Assessment of ultrafine particles and noise measurements using fuzzy logic and data mining techniques.

This study focuses on correlations between total number concentrations, road traffic emissions and noise levels in an urban area in the southwest of Spain during the winter and summer of 2009. The high temporal correlation between sound pressure levels, traffic intensity, particle number concentrations related to traffic, black carbon and NOx concentrations suggests that noise is linked to traffic emissions as a main source of pollution in urban areas. First, the association of these different variables was studied using PreFuRGe, a computational tool based on data mining and fuzzy logic. The results showed a clear association between noise levels and road-traffic intensity for non-extremely high wind speed levels. This behaviour points, therefore, to vehicular emissions being the main source of urban noise. An analysis for estimating the total number concentration from noise levels is also proposed in the study. The high linearity observed between particle number concentrations linked to traffic and noise levels with road traffic intensity can be used to calculate traffic related particle number concentrations experimentally. At low wind speeds, there are increases in noise levels of 1 dB for every 100 vehicles in circulation. This is equivalent to 2000 cm(-3) per vehicle in winter and 500 cm(-3) in summer. At high wind speeds, wind speed could be taken into account. This methodology allows low cost sensors to be used as a proxy for total number concentration monitoring in urban air quality networks.

COMPUTER-AIDED DIAGNOSIS OF THE PAROXYSMAL ATRIAL FIBRILLATION: A FUZZY-EVOLUTIONARY APPROACH

This paper presents an integrated fuzzy-evolutionary methodology to address computer-aided diagnosis in medical applications by using features extracted from biosignal processing. In the proposed methodology, a deterministic crowding genetic optimizer designed to provide high-diversity solutions is used for weighted feature selection, and the diagnostic decision is made by a binary k-nearest neighbor classifier. Weight vector solutions resulting from the optimization stage are processed by a fuzzy rules generator to retrieve a fuzzy model giving friendly information to the medical specialist about the role of the different features in the diagnosis. This allows the design of efficient diagnosis protocols. The overall diagnostic methodology is applied to Paroxysmal Atrial Fibrillation detection based on analysis of nonfibrillation ECGs, obtaining a fuzzy model consistent with previous work in this field.

Behavior patterns in hormonal treatments using fuzzy logic models

Assisted reproductive technologies are a combination of medical strategies designed to treat infertility patients. Ideal stimulation treatment has to be individualized, but one of the main challenges which clinicians face in the everyday clinic is how to select the best medical protocol for a patient. This work aims to look for behavior patterns in this kind of treatments, using fuzzy logic models with the objective of helping gynecologists and embryologists to make decisions that could improve the process of in vitro fertilization. For this purpose, a real-world dataset composed of one hundred and twenty-three (123) patients and five hundred and fifty-nine (559) treatments applied in relation to such patients provided by an assisted reproduction clinic, has been used to obtain the fuzzy models. As conclusion, this work corroborates some known clinic experiences, provides some new ones and proposes a set of questions to be solved in future experiments.

A methodology to generate compact and accurate fuzzy knowledge bases based on fuzzy clustering and evolutionary selection and tuning

A new methodology to learn descriptive linguistic Fuzzy Rule-based System Knowledge Bases from examples based on the combination of fuzzy clustering and evolutionary simultaneous rule selection and membership functions tuning is presented in this work. Fuzzy clustering is used to achieve a preliminary description of the data, in other words to obtain information on the definition of the linguistic terms and rules instead of predefined linguistic terms and rules that use them. The evolutionary algorithm obtains the final compact and accurate knowledge base selecting a subset of rules with high level of cooperation and fine-tuning the linguistic terms involved. The results obtained with this proposal improves accuracy as well as complexity through the number of rules compared with a classic algorithm and a reference algorithm both well known in the literature, as the experimental study developed shows, using several different data sets.

Fuzzy-Genetic Methodology for Web-based Computed-Aided Diagnosis in Medical Applications

This paper presents an integrated fuzzy-genetic methodology to address web-based computed-aided diagnosis by using bio-signal processing in medical applications. A deterministic crowding genetic algorithm is used for obtaining different subsets of features that provide high performance classification in a K-Nearest Neighbor classifier. These subsets of features are then used as training data in a rule generator based on fuzzy clustering to obtain a performance qualitative model that can give information about the more suitable features to use in the diagnosis. This model can also be used to assess the (performance) accuracy that will be reached by using a given set of features – possible those ones available at a specific medical centre. The overall methodology is applied to Paroxysmal Atrial Fibrillation (PAF) – the heart arrhythmia that causes more frequently cerebrovascular incidents – Diagnosis based on analysis of nonfibrillation ECGs