Kernel machines for epilepsy diagnosis via EEG signal classification: A comparative study

Objective: We carry out a systematic assessment on a suite of kernel-based learning machines while coping with the task of epilepsy diagnosis through automatic electroencephalogram (EEG) signal classification. Methods and materials: The kernel machines investigated include the standard support vector machine (SVM), the least squares SVM, the Lagrangian SVM, the smooth SVM, the proximal SVM, and the relevance vector machine. An extensive series of experiments was conducted on publicly available data, whose clinical EEG recordings were obtained from five normal subjects and five epileptic patients. The performance levels delivered by the different kernel machines are contrasted in terms of the criteria of predictive accuracy, sensitivity to the kernel function/parameter value, and sensitivity to the type of features extracted from the signal. For this purpose, 26 values for the kernel parameter (radius) of two well-known kernel functions (namely. Gaussian and exponential radial basis functions) were considered as well as 21 types of features extracted from the EEG signal, including statistical values derived from the discrete wavelet transform, Lyapunov exponents, and combinations thereof. Results: We first quantitatively assess the impact of the choice of the wavelet basis on the quality of the features extracted. Four wavelet basis functions were considered in this study. Then, we provide the average accuracy (i.e., cross-validation error) values delivered by 252 kernel machine configurations; in particular, 40%/35% of the best-calibrated models of the standard and least squares SVMs reached 100% accuracy rate for the two kernel functions considered. Moreover, we show the sensitivity profiles exhibited by a large sample of the configurations whereby one can visually inspect their levels of sensitiveness to the type of feature and to the kernel function/parameter value. Conclusions: Overall, the results evidence that all kernel machines are competitive in terms of accuracy, with the standard and least squares SVMs prevailing more consistently. Moreover, the choice of the kernel function and parameter value as well as the choice of the feature extractor are critical decisions to be taken, albeit the choice of the wavelet family seems not to be so relevant. Also, the statistical values calculated over the Lyapunov exponents were good sources of signal representation, but not as informative as their wavelet counterparts. Finally, a typical sensitivity profile has emerged among all types of machines, involving some regions of stability separated by zones of sharp variation, with some kernel parameter values clearly associated with better accuracy rates (zones of optimality). (C) 2011 Elsevier B.V. All rights reserved.

Multilabeled classification approach to find a plaint source for terpenoids

Recently, we have built a classification model that is capable of assigning a given sesquiterpene lactone (STL) into exactly one tribe of the plant family Asteraceae from which the STL has been isolated. Although many plant species are able to biosynthesize a set of peculiar compounds, the occurrence of the same secondary metabolites in more than one tribe of Asteraceae is frequent. Building on our previous work, in this paper, we explore the possibility of assigning an STL to more than one tribe (class) simultaneously. When an object may belong to more than one class simultaneously, it is called multilabeled. In this work, we present a general overview of the techniques available to examine multilabeled data. The problem of evaluating the performance of a multilabeled classifier is discussed. Two particular multilabeled classification methods-cross-training with support vector machines (ct-SVM) and multilabeled k-nearest neighbors (M-L-kNN)were applied to the classification of the STLs into seven tribes from the plant family Asteraceae. The results are compared to a single-label classification and are analyzed from a chemotaxonomic point of view. The multilabeled approach allowed us to (1) model the reality as closely as possible, (2) improve our understanding of the relationship between the secondary metabolite profiles of different Asteraceae tribes, and (3) significantly decrease the number of plant sources to be considered for finding a certain STL. The presented classification models are useful for the targeted collection of plants with the objective of finding plant sources of natural compounds that are biologically active or possess other specific properties of interest.

Selection bias in gene extraction on the basis of microarray gene-expression data

In the context of cancer diagnosis and treatment, we consider the problem of constructing an accurate prediction rule on the basis of a relatively small number of tumor tissue samples of known type containing the expression data on very many (possibly thousands) genes. Recently, results have been presented in the literature suggesting that it is possible to construct a prediction rule from only a few genes such that it has a negligible prediction error rate. However, in these results the test error or the leave-one-out cross-validated error is calculated without allowance for the selection bias. There is no allowance because the rule is either tested on tissue samples that were used in the first instance to select the genes being used in the rule or because the cross-validation of the rule is not external to the selection process; that is, gene selection is not performed in training the rule at each stage of the cross-validation process. We describe how in practice the selection bias can be assessed and corrected for by either performing a cross-validation or applying the bootstrap external to the selection process. We recommend using 10-fold rather than leave-one-out cross-validation, and concerning the bootstrap, we suggest using the so-called. 632+ bootstrap error estimate designed to handle overfitted prediction rules. Using two published data sets, we demonstrate that when correction is made for the selection bias, the cross-validated error is no longer zero for a subset of only a few genes.

A framework for electricity price spike analysis with advanced data mining methods

There are many techniques for electricity market price forecasting. However, most of them are designed for expected price analysis rather than price spike forecasting. An effective method of predicting the occurrence of spikes has not yet been observed in the literature so far. In this paper, a data mining based approach is presented to give a reliable forecast of the occurrence of price spikes. Combined with the spike value prediction techniques developed by the same authors, the proposed approach aims at providing a comprehensive tool for price spike forecasting. In this paper, feature selection techniques are firstly described to identify the attributes relevant to the occurrence of spikes. A simple introduction to the classification techniques is given for completeness. Two algorithms: support vector machine and probability classifier are chosen to be the spike occurrence predictors and are discussed in details. Realistic market data are used to test the proposed model with promising results.

Improving Alzheimer`s Disease Diagnosis with Machine Learning Techniques

There is not a specific test to diagnose Alzheimer`s disease (AD). Its diagnosis should be based upon clinical history, neuropsychological and laboratory tests, neuroimaging and electroencephalography (EEG). Therefore, new approaches are necessary to enable earlier and more accurate diagnosis and to follow treatment results. In this study we used a Machine Learning (ML) technique, named Support Vector Machine (SVM), to search patterns in EEG epochs to differentiate AD patients from controls. As a result, we developed a quantitative EEG (qEEG) processing method for automatic differentiation of patients with AD from normal individuals, as a complement to the diagnosis of probable dementia. We studied EEGs from 19 normal subjects (14 females/5 males, mean age 71.6 years) and 16 probable mild to moderate symptoms AD patients (14 females/2 males, mean age 73.4 years. The results obtained from analysis of EEG epochs were accuracy 79.9% and sensitivity 83.2%. The analysis considering the diagnosis of each individual patient reached 87.0% accuracy and 91.7% sensitivity.

Use of SVM Methods with Surface-Based Cortical and Volumetric Subcortical Measurements to Detect Alzheimer`s Disease

Here, we examine morphological changes in cortical thickness of patients with Alzheimer`s disease (AD) using image analysis algorithms for brain structure segmentation and study automatic classification of AD patients using cortical and volumetric data. Cortical thickness of AD patients (n = 14) was measured using MRI cortical surface-based analysis and compared with healthy subjects (n = 20). Data was analyzed using an automated algorithm for tissue segmentation and classification. A Support Vector Machine (SVM) was applied over the volumetric measurements of subcortical and cortical structures to separate AD patients from controls. The group analysis showed cortical thickness reduction in the superior temporal lobe, parahippocampal gyrus, and enthorhinal cortex in both hemispheres. We also found cortical thinning in the isthmus of cingulate gyrus and middle temporal gyrus at the right hemisphere, as well as a reduction of the cortical mantle in areas previously shown to be associated with AD. We also confirmed that automatic classification algorithms (SVM) could be helpful to distinguish AD patients from healthy controls. Moreover, the same areas implicated in the pathogenesis of AD were the main parameters driving the classification algorithm. While the patient sample used in this study was relatively small, we expect that using a database of regional volumes derived from MRI scans of a large number of subjects will increase the SVM power of AD patient identification.

Evaluating SVM and MLDA in the extraction of discriminant regions for mental state prediction

Pattern recognition methods have been successfully applied in several functional neuroimaging studies. These methods can be used to infer cognitive states, so-called brain decoding. Using such approaches, it is possible to predict the mental state of a subject or a stimulus class by analyzing the spatial distribution of neural responses. In addition it is possible to identify the regions of the brain containing the information that underlies the classification. The Support Vector Machine (SVM) is one of the most popular methods used to carry out this type of analysis. The aim of the current study is the evaluation of SVM and Maximum uncertainty Linear Discrimination Analysis (MLDA) in extracting the voxels containing discriminative information for the prediction of mental states. The comparison has been carried out using fMRI data from 41 healthy control subjects who participated in two experiments, one involving visual-auditory stimulation and the other based on bimanual fingertapping sequences. The results suggest that MLDA uses significantly more voxels containing discriminative information (related to different experimental conditions) to classify the data. On the other hand, SVM is more parsimonious and uses less voxels to achieve similar classification accuracies. In conclusion, MLDA is mostly focused on extracting all discriminative information available, while SVM extracts the information which is sufficient for classification. (C) 2009 Elsevier Inc. All rights reserved.

Análise do nível de estresse baseada em sinais de eletroencefalografia e de condutância da pele

O estresse pode afetar qualquer pessoa, independente de idade, sexo ou etnia. O organismo humano o utiliza como uma resposta adaptativa frente a situações diversas, as quais requeiram alguma adaptação do organismo para que possa enfrentar tal situação. Dependendo do estímulo estressor, pode ser gerado no indivíduo desgastes físico, mental ou emocional, no entanto, o estresse não representa necessariamente algo ruim ou patológico; este é um mecanismo de adaptação vital para a sobrevivência da espécie humana. Porém, o número de pessoas que são afetadas de forma negativa pelo estresse tem crescido imensamente nas últimas décadas. Pesquisas destacam que nos Estados Unidos cerca de 60% a 90% dos atendimentos médicos estão relacionados de alguma maneira com o estresse, enquanto que no Brasil aproximadamente 80% da população sofre de estresse, sendo que desses, 30% encontram-se na fase mais crítica, a chamada fase de exaustão. Tendo em vista que a principal forma de identificação de estresse ainda é realizada através do uso de questionário de autorrelato. O presente estudo apresenta como contribuição uma metodologia de análise do nível de estresse baseada na variação da condutância galvânica da pele e de sinais de eletroencefalografia, sendo utilizados como parâmetros a assimetria do ritmo alfa, assim como a razão entre os ritmos beta e alfa no córtex frontal e pré-frontal. Para a gravação dos sinais de EEG foi utilizado um dispositivo portátil, com eletrodos especificamente situados nas posições aF3, F3, F4 e aF4, de acordo com o Sistema Internacional 10/20 de posicionamento de eletrodos. Os participantes deste estudo são Bombeiros Militares da 1ª Cia de Vitória-ES. Foram utilizadas três classes de estímulos emocionais positivos, calmos e negativos, através da utilização de imagens pertencentes ao banco de dados IAPS (International Affective Picture System). Os resultados de acurácia obtidos através de um classificador SVM (Support Vector Machine) chegam a 88,24% para classe de estímulos positivos, 84,09% para classe calma e de 92,86% para os estímulos negativos. Deste modo, esta pesquisa apresenta uma combinação de parâmetros que podem ser aferidos com equipamentos de baixo custo, e fornecem condições de diferenciar estímulos estressantes, podendo assim, ser utilizada para auxiliar no treinamento de profissionais da área de urgência e emergência.

Towards Emotional Interaction: Using Movies to Automatically Learn Users' Emotional States

The HCI community is actively seeking novel methodologies to gain insight into the user’s experience during interaction with both the application and the content. We propose an emotional recognition engine capable of automatically recognizing a set of human emotional states using psychophysiological measures of the autonomous nervous system, including galvanic skin response, respiration, and heart rate. A novel pattern recognition system, based on discriminant analysis and support vector machine classifiers is trained using movies’ scenes selected to induce emotions ranging from the positive to the negative valence dimension, including happiness, anger, disgust, sadness, and fear. In this paper we introduce an emotion recognition system and evaluate its accuracy by presenting the results of an experiment conducted with three physiologic sensors.

DESENVOLVIMENTO DE UM CLASSIFICADOR PARA ANÁLISE BIOMÉTRICA DE UM COLETE INSTRUMENTADO PARA REEDUCAÇÃO POSTURAL

Ao longo dos tempos tem existido um avanço, nas empresas, dirigido à preocupação com o bemestar dos trabalhadores, adotando por isso medidas preventivas. A formação especializada em Medicina do Trabalho é indispensável para o exercício de atividades de prevenção dos riscos profissionais e de promoção da saúde. A postura corporal pode ser definida como a posição e a orientação global do corpo e membros relativamente uns aos outros. Qualquer desvio na forma da coluna vertebral pode gerar solicitações funcionais prejudiciais que ocasionam um aumento de fadiga no trabalhador e leva ao longo do tempo a lesões graves. Cada vez mais surgem doenças profissionais provocadas pela adoção de más posturas, na realização de tarefas diárias dos trabalhadores. A boa postura corporal é uma tarefa específica que representa uma interação complexa entre a função biomecânica e neuromuscular. No presente plano de dissertação foram estudados diferentes classificadores tendo como objetivo classificar boas e más posturas corporais de trabalhadores em contexto de trabalho. Assim foram estudados diferentes classificadores de machine learnig, redes neuronais artificiais, support vector machine, árvores de decisão, análise discriminante, regressão logística, treebagger e naíve bayes. Para treino de classificadores foi realizada a aquisição tridimensional da postura da espinha a 100 pessoas, passando por uma parametrização e treino de diferentes classificadores para a determinação automática do tipo de postura corporal. O classificador que obteve melhor desempenho foi o Treebagger com uma classificação para True Positive de 93,3% e True Negative de 96,2%.

Agrupamento multi-orador

Actualmente tem-se observado um aumento do volume de sinais de fala em diversas aplicações, que reforçam a necessidade de um processamento automático dos ficheiros. No campo do processamento automático destacam-se as aplicações de “diarização de orador”, que permitem catalogar os ficheiros de fala com a identidade de oradores e limites temporais de fala de cada um, através de um processo de segmentação e agrupamento. No contexto de agrupamento, este trabalho visa dar continuidade ao trabalho intitulado “Detecção do Orador”, com o desenvolvimento de um algoritmo de “agrupamento multi-orador” capaz de identificar e agrupar correctamente os oradores, sem conhecimento prévio do número ou da identidade dos oradores presentes no ficheiro de fala. O sistema utiliza os coeficientes “Mel Line Spectrum Frequencies” (MLSF) como característica acústica de fala, uma segmentação de fala baseada na energia e uma estrutura do tipo “Universal Background Model - Gaussian Mixture Model” (UBM-GMM) adaptado com o classificador “Support Vector Machine” (SVM). No trabalho foram analisadas três métricas de discriminação dos modelos SVM e a avaliação dos resultados foi feita através da taxa de erro “Speaker Error Rate” (SER), que quantifica percentualmente o número de segmentos “fala” mal classificados. O algoritmo implementado foi ajustado às características da língua portuguesa através de um corpus com 14 ficheiros de treino e 30 ficheiros de teste. Os ficheiros de treino dos modelos e classificação final, enquanto os ficheiros de foram utilizados para avaliar o desempenho do algoritmo. A interacção com o algoritmo foi dinamizada com a criação de uma interface gráfica que permite receber o ficheiro de teste, processá-lo, listar os resultados ou gerar um vídeo para o utilizador confrontar o sinal de fala com os resultados de classificação.

Modelo de representação de texto mais adequado à classificação

Mestrado em Engenharia Informática

Classification and staging of chronic liver disease from multimodal data

Chronic liver disease (CLD) is most of the time an asymptomatic, progressive, and ultimately potentially fatal disease. In this study, an automatic hierarchical procedure to stage CLD using ultrasound images, laboratory tests, and clinical records are described. The first stage of the proposed method, called clinical based classifier (CBC), discriminates healthy from pathologic conditions. When nonhealthy conditions are detected, the method refines the results in three exclusive pathologies in a hierarchical basis: 1) chronic hepatitis; 2) compensated cirrhosis; and 3) decompensated cirrhosis. The features used as well as the classifiers (Bayes, Parzen, support vector machine, and k-nearest neighbor) are optimally selected for each stage. A large multimodal feature database was specifically built for this study containing 30 chronic hepatitis cases, 34 compensated cirrhosis cases, and 36 decompensated cirrhosis cases, all validated after histopathologic analysis by liver biopsy. The CBC classification scheme outperformed the nonhierachical one against all scheme, achieving an overall accuracy of 98.67% for the normal detector, 87.45% for the chronic hepatitis detector, and 95.71% for the cirrhosis detector.

The usefulness of ultrasound in the classification of chronic liver disease

Chronic Liver Disease is a progressive, most of the time asymptomatic, and potentially fatal disease. In this paper, a semi-automatic procedure to stage this disease is proposed based on ultrasound liver images, clinical and laboratorial data. In the core of the algorithm two classifiers are used: a k nearest neighbor and a Support Vector Machine, with different kernels. The classifiers were trained with the proposed multi-modal feature set and the results obtained were compared with the laboratorial and clinical feature set. The results showed that using ultrasound based features, in association with laboratorial and clinical features, improve the classification accuracy. The support vector machine, polynomial kernel, outperformed the others classifiers in every class studied. For the Normal class we achieved 100% accuracy, for the chronic hepatitis with cirrhosis 73.08%, for compensated cirrhosis 59.26% and for decompensated cirrhosis 91.67%.

Chronic liver disease staging classification based on ultrasound, clinical and laboratorial data

In this work the identification and diagnosis of various stages of chronic liver disease is addressed. The classification results of a support vector machine, a decision tree and a k-nearest neighbor classifier are compared. Ultrasound image intensity and textural features are jointly used with clinical and laboratorial data in the staging process. The classifiers training is performed by using a population of 97 patients at six different stages of chronic liver disease and a leave-one-out cross-validation strategy. The best results are obtained using the support vector machine with a radial-basis kernel, with 73.20% of overall accuracy. The good performance of the method is a promising indicator that it can be used, in a non invasive way, to provide reliable information about the chronic liver disease staging.