L.J. Herrera

Multiwindow Fusion for Wearable Activity Recognition

The recognition of human activity has been extensively investigated in the last decades. Typically, wearable sensors are used to register body motion signals that are analyzed by following a set of signal processing and machine learning steps to recognize the activity performed by the user. One of the most important steps refers to the signal segmentation, which is mainly performed through windowing approaches. In fact, it has been proved that the choice of window size directly conditions the performance of the recognition system. Thus, instead of limiting to a specific window configuration, this work proposes the use of multiple recognition systems operating on multiple window sizes. The suggested model employs a weighted decision fusion mechanism to fairly leverage the potential yielded by each recognition system based on the target activity set. This novel technique is benchmarked on a well-known activity recognition dataset. The obtained results show a significant improvement in terms of performance with respect to common systems operating on a single window size.

Effect of pre-processing methods on microarray-based SVM classifiers in affymetrix genechips

Affymetrix High Oligonucleotide expression arrays are widely used for the high-throughput assessment of gene expression of thousands of genes simultaneously. Although disputed by several authors, there are non-biological variations and systematic biases that must be removed as much as possible through the pre-processing step before an absolute expression level for every gene is assessed. It is important to evaluate microarray pre-processing procedures not only to the detection of differentially expressed genes, but also to classification, since a major use of microarrays is the expression-based phenotype classification. Thus, in this paper, we use several cancer microarray datasets to assess the influence of five different pre-processing methods in Support Vector Machine-based classification methodologies with different kernels: linear, Radial Basis Functions (RBFs) and polynomial.

Intelligent Systems to Autonomously Classify Several Arrhythmia Using Information from ECG

This paper is focused on the development of intelligent classifiers in the area of biomedicine, focusing on the problem of diagnosing cardiac diseases based on the electrocardiogram (ECG), or more precisely, on the differentiation of several arrhythmia using a large data set, by an autonomous intelligent system which can be used as an expert system to support human experts in the diagnosis and, moreover, to autonomously display an alarm to the user in case of a dangerous situation. We will study and imitate the ECG treatment methodologies and the features extracted from the electrocardiograms used by the researchers, which obtained the best results in the PhysioNet Challenge. We will extract a great amount of features, partly those used by these researchers and some additional others we considered to be important for the distinction previously mentioned. A new method based on different paradigms of intelligent computation (such as extreme learning machine, support vector machine and feature selection) will be used to select the most relevant characteristics and to obtain a classifier capable of autonomously distinguishing the different types arrhythmia from the ECG signal. Finally, the behavior and performance of the classifier have been tested using data from several cardiac pathologies, obtaining good classification results.

Comparison of different computational intelligent classifier to autonomously detect cardiac pathologies diagnosed by ECG

The electrocardiogram (ECG) is a noninvasive technique used to reflect underlying heart conditions by measuring the electrical activity of the heart, and nowadays it is possible with just a few derivation (with just only two), obtain important information in order than an expert can recognize abnormal heart rhythms (the heart rate is very fast, very slow, or irregular) or a heart attack (myocardial infarction), and if it was recent or some time ago. In this paper, several intelligent classifiers are developed, focusing on the problem of diagnosing cardiac diseases based on the electrocardiogram (ECG), or more precisely, on the differentiation of several arrhythmia using a large data set. We will study and imitate the ECG treatment methodologies and the features extracted from the electrocardiograms used by the researchers, which obtained the best results in the PhysioNet Challenge (www.physionet.org/). We will extract a great amount of features, partly those used by these researchers and some additional others we considered to be important for the distinction previously mentioned. A new method based on different paradigms of intelligent computation (such as extreme learning machine, support vector machine, decision trees, genetic algorithms and feature selection) will be used to select the most relevant characteristics and to obtain a classifier capable of autonomously distinguishing the different types arrhythmia from the ECG signal.

Applying support vector machines and mutual information to book losses prediction

This work presents a feasible solution to the problem of book losses prediction from financial and general data in companies. The specific problem tackled in this work corresponds to a real dataset of Spanish companies. A Mutual Information-based criterion has been applied in order to reduce the initial set of variables, and a Support Vector Machine classifier has been designed to perform the prediction. The results show that the proposed approach obtains an important reduction of the number of variables needed to perform the prediction, improving the generalization capabilities of the model. The accuracy rates were above the 84% in the test set, much better than those obtained by other soft-computing algorithms (such as Genetic Programming, Self-Organizing Maps or Artificial Neural Networks) working with the same dataset and presented in previous works. The proposed approach shows to be promising and could be determinant in providing the experts with the right tools for the selection of the relevant factors and for the prediction in this difficult problem.

A Data Mining Approach Based on a Local-Global Fuzzy Modelling for Prediction of Color Change after Tooth Bleaching Using Vita Classical Shades

Tooth bleaching is receiving an increasing interest by patients and dentists since it is a relatively non-invasive approach for whitening and lightening teeth. Instrument designed for tooth color measurements and visual assessment with commercial shade guides are nowadays used to evaluate the tooth color. However, the degree of color change after tooth bleaching varied substantially among studies and currently, there are no objective guidelines to predict the effectiveness of a tooth bleaching treatment. Fuzzy Logic is a well known paradigm for data modelling; their main advantage is their ability to provide an interpretable set of rules that can be later used by the scientists. However these models have the problem that the global approximation optimization can lead to a deficient rule local modelling. This work proposes a modified fuzzy model that performs a simultaneous global and local modelling. This property is reached thanks to a special partitioning of the input space in the fuzzy system. The proposed approach is used to approximate a set of color measurements taken after a bleaching treatment using the pre-bleaching measurements. The system uses as rule antecedents the colorimetric values of the VITA commercial shade guide. The expected post-bleaching colorimetric values are immediately obtained from the local models (rules) of the system thanks to the proposed modified fuzzy model. Additionally, these post-bleaching CIELAB coordinate values have been associated with VITA shades through the evaluation of their respective membership functions, approximating which VITA shades are expected after the treatment for each pre-bleaching VITA shade.

Self-adaptive robot control using fuzzy logic

This paper presents a new methodology to achieve real time self tuning and self learning in fuzzy controllers, with application to motion control of a trailer for reaching an aiming point and obstacle-avoidance. The advantage of this approach is that it only requires qualitative information about the plant to be controlled, in terms of the monotony presented by the output with respect to the control signal and the delay of the plant. Also, starting with a non-optimum controller, the system is able to self-adapt its behaviour in order to reduce the error. Thus, it is capable of controlling highly non-linear systems, in a pseudo-optimum way, even when these are time variable, for example, the dynamic of the robot or trailer change (i.e.: different mass, different environments, different dynamics of the system). Control is achieved by means of two auxiliary systems: the first one is responsible for adapting the consequences of the main controller to minimize the error arising at the plant output, while the second auxiliary system compiles real input/output data obtained from the plant. The methodology has been successfully applied to a real robot with different dynamics and in different environments, showing its ability to tune its +behaviour

Performing Variable Selection by Multiobjective Criterion: An Application to Mobile Payment

The rapid growth social networks have led many companies to use mobile payment systems as business sales tools. As these platforms have an increasing acceptance among the consumers, the main goal of this research is to analyze the individuals’ use intention of these systems in a social network environment. The problem of variable selection arises in this context as key to understand user’s behaviour. This paper compares several non-parametric criteria to perform variable selection and combines them in a multiobjective manner showing a good performance in the experiments carried out and validated by experts.

Modeling the EUR/USD Index Using LS-SVM and Performing Variable Selection

As machine learning becomes more popular in all fields, its use is well known in finance and economics. The growing number of people using models to predict the market’s behaviour can modify the market itself so it is more predictable. In this context, the key element is to find out which variables are used to build the model in a macroeconomic environment. This paper presents an application of kernel methods to predict the EUR/USD relationship performing variable selection. The results show how after applying a proper variable selection, very accurate predictions can be achieved and smaller historical data is needed to train the model.

Color Fuzzy Set Design for dental applications

This work presents a Color Fuzzy Set Design procedure, or similarly a color naming procedure, based on the VITA Classical dental color shade guide, for composite color matching applications in direct restorations in aesthetic dentistry. For that purpose, a psychophysical experience based on a set of visual assessments by a panel of observers was performed, with the objective of identifying the areas of the CIELAB color space which can be associated to each shade of the VITA guide. A number of composite sets from four different manufacturing companies were used in the experiment. This information was processed and a set of nonparametric Fuzzy Sets was obtained for the VITA shades, with the aid of an acceptability dental color threshold taken from previous studies. The designed Fuzzy Set System was applied for the identification of the VITA shade of different dentine and enamel composite samples (bilaminar technique) from two different manufacturing companies, with the subsequent clinical applicability. Results show a high level of accuracy in VITA shade identification of the operated dental samples, when compared with visual assessment performed by experts.