Cardiotocographic (CTG) monitoring is a method of assessing fetal state. Since visual analysis of CTG signal is difficult, methods of automated qualitative fetal state evaluation on the basis of the quantitative description of the signal are applied. The appropriate selection of learning data influences the quality of the fetal state assessment with computational intelligence methods. In the presented work we examined three different feature selection procedures based on: principal components analysis, receiver operating characteristics and guidelines of International Federation of Gynecology and Obstetrics. To investigate their influence on the fetal state assessment quality the benchmark SisPorto® dataset and the Lagrangian support vector machine were used.
Recording and analysis of fetal heart rate (FHR) signal is nowadays the primary method for the biophysical assessment of the fetal state. Since the correct interpretation of crucial FHR characteristics is difficult, methods of automated quantitative signal evaluation are still the subject of the research studies. In the following paper we investigated the possibility of improvement of the fetal state evaluation on the basis of the epsilon-insensitive learning (eIL). We examined two eIL procedures integrated with fuzzy clustering algorithms as well as different methods of logical interpretation of the fuzzy conditional statements. The quality of the FHR signal classification was evaluated using the data collected with the computerized fetal surveillance system. The learning performance was measured with the number of correct classification (CC) and overall quality index (QI) defined as a geometric mean of sensitivity and specificity. The obtained results (CC = 88 % and QI = 87 %) show a high efficiency of the fetal state assessment using the epsilon-insensitive learning based methods.
Classification plays an important role in many fields of life, including medical diagnosis support. In the paper, fuzzy clustering algorithm dedicated to classification methods is proposed. Its goal is to find pairs of prototypes located near boundaries of both classes of objects. The minimization procedure of the proposed criterion function is described. The algorithm for determining the value of the clustering parameter is also presented. Presented results (synthetic dataset) confirm correctness of clustering - most of final prototypes, determined based on obtained pairs, are located between boundary of two classes.
Linear regression analysis has become a fundamental tool in experimental sciences. We propose a new method for parameter estimation in linear models. The 'Generalized Ordered Linear Regression with Regularization' (GOLRR) uses various loss functions (including the o-insensitive ones), ordered weighted averaging of the residuals, and regularization. The algorithm consists in solving a sequence of weighted quadratic minimization problems where the weights used for the next iteration depend not only on the values but also on the order of the model residuals obtained for the current iteration. Such regression problem may be transformed into the iterative reweighted least squares scenario. The conjugate gradient algorithm is used to minimize the proposed criterion function. Finally, numerical examples are given to demonstrate the validity of the method proposed.
W pracy przedstawiono ujednolicenie metod opisu sygnałów elektrokardiograficznych za pomocą koncepcji szeregu czasowego zbiorów rozmytych. Sygnał elektrokardiograficzny jest przetwarzany w "ruchomym oknie" czasowym i przekształcany na szereg czasowy funkcji przynależności zbiorów rozmytych. Jako szczególne przypadki wprowadzonej metody można rozważać koncepcję sygnału rozmytego oraz ciąg ziaren informacyjnych.
EN
The unification of the electrocardiography signals description methods by means of time series of fuzzy sets, is presented. The electrocardiographic signal is processed in a “moving window” and transformed into a time series of fuzzy sets membership functions. The idea of fuzzy signal and the sequence of information granules can be considered as special case of the presented method.
This paper introduces a new classifier design method based on regularized iteratively reweighted least squares criterion function. The proposed method uses various approximations of misclassification error, including: linear, sigmoidal, Huber and logarithmic. Using the represented theorem a kernel version of classifier design method is introduced. The conjugate gradient algorithm is used to minimize the proposed criterion function. Furthermore, .1-regularized kernel version of the classifier is introduced. In this case, the gradient projection is used to optimize the criterion function. Finally, an extensive experimental analysis on 14 benchmark datasets is given to demonstrate the validity of the introduced methods.
Cardiotocography (CTG) is a routine method of fetal condition assessment used in modern obstetrics. It is a biophysical method based on simultaneous recording and analysis of activity of fetal heart, fetal movements and maternal uterine contractions. The fetal condition is diagnosed on the basis of printed CTG trace evaluation. The correct interpretation of CTG traces from a bedside monitor is very difficult even for experienced clinicians. Therefore, computerized fetal monitoring systems are used to yield the quantitative description of the signal. However, the effective methods, aiming to support the conclusion generation, are still being searched. One of the most important features defining the state of fetal outcome is the weight of the newborn. The presented work describes an application of the Artificial Neural Network Based on Logical Interpretation of fuzzy if-then Rules (ANBLIR) to evaluate the risk of the low birth weight using a set of parameters quantitatively describing the CTG traces. The obtained results confirm that the neuro-fuzzy based CTG classification methods are very efficient for the prediction of the fetal outcome.
Cardiotocographic monitoring (CTG) is a primary biophysical monitoring method for assessment of the fetal state and is based on analysis of fetal heart rate, uterine contraction activity and fetal movement signals. Visual analysis of CTG traces is very difficult so computer-aided fetal monitoring systems have become a standard in clinical centres. We proposed the application of neural networks for the prediction of fetal outcome using the parameters of quantitative description of acquired signals as inputs. We focused on the influence of the gestational age (during trace recording) on the fetal outcome classification quality. We designed MLP and RBF neural networks with changing the number of neurons in the hidden layer to find the best structure. Networks were trained and tested fifty times, with random cases assignment to training, validating and testing subset. We obtained the value of sensitivity index above 0.7, what may be regarded as good result. However additional trace grouping within similar gestational age, increased classification quality in the case of MLP networks.
The paper presents new approach to problem of signal averaging which is commonly used to extract a useful signal distorted by a noise. The averaging is especially useful for biomedical signal such as ECG signal, where the spectra of the signal and noise significantly overlap. In reality can be observed variability of noise power from cycle to cycle which is motivation for using methods of weighted averaging. Performance of the new method, based on partition of input set in time domain and criterion function minimization, is experimentally compared with the traditional averaging by using arithmetic mean, weighted averaging method based on empirical Bayesian approach and weighted averaging method based on criterion function minimization.
10
Dostęp do pełnego tekstu na zewnętrznej witrynie WWW
One of the prime tool in non-invasive cardiac electrophysiology is the recording of an electrocardiographic signal (ECG) which analysis is greatly useful in the screening and diagnosis of cardiovascular diseases. However, one of the greatest problems is that usually recording an electrical activity of the heart is performed in the presence of noise. The paper presents Bayesian and empirical Bayesian approach to problem of weighted signal averaging in time domain which is commonly used to extract a useful signal distorted by a noise. The averaging is especially useful for biomedical signal such as ECG signal, where the spectra of the signal and noise significantly overlap. Using the methods of weighted averaging are motivated by variability of noise power from cycle to cycle, often observed in reality. It is demonstrated that exploiting a probabilistic Bayesian learning framework leads to accurate prediction models. Additionally, even in the presence of nuisance parameters the empirical Bayesian approach offers the method of theirs automatic estimation which reduces number of preset parameters. Performance of the new method is experimentally compared to the traditional averaging by using arithmetic mean and weighted averaging method based on criterion function minimization.
An electrocardiogram (ECG) is the prime tool in non-invasive cardiac electrophysiology and has a prime function in the screening and diagnosis of cardiovascular diseases. However one of the greatest problems is that usually recording an electrical activity of the heart is performed in the presence of noise. The paper presents empirical Bayesian approach to problem of signal averaging which is commonly used to extract a useful signal distorted by a noise. The averaging is especially useful for biomedical signal such as ECG signal, where the spectra of the signal and noise significantly overlap. In reality the variability of noise can be observed, with power from cycle to cycle, which is motivation for weighted averaging methods usage. It is demonstrated that by exploiting a probabilistic Bayesian learning framework, it can be derived accurate prediction models offering significant additional advantage, namely automatic estimation of 'nuisance' parameters. Performance of the new method is experimentally compared to the traditional averaging by using arithmetic mean and weighted averaging method based on criterion function minimization.
The paper presents new method called the Fuzzy Relevance Vector Machine (FRVM), a modification of the relevance vector machine, introduced by M. Tipping, applied to learning Takagi-Sugeno-Kang (TSK) fuzzy system. Moreover it describes application of the FRVM to noise reduction in ECG signal. The results of the process are compared to those obtained using both Least Squares method for learning output functions in TSK rules and commonly used method using a low-pass moving average filter.
13
Dostęp do pełnego tekstu na zewnętrznej witrynie WWW
This paper introduces a new classifier design method based on a kernel extension of the classical Ho-Kashyap procedure. The proposed method uses an approximation of the absolute error rather than the squared error to design a classifier, which leads to robustness against outliers and a better approximation of the misclassification error. Additionally, easy control of the generalization ability is obtained using the structural risk minimization induction principle from statistical learning theory. Finally, examples are given to demonstrate the validity of the introduced method.
14
Dostęp do pełnego tekstu na zewnętrznej witrynie WWW
This paper introduces a new classifier design method that is based on a modification of the classical Ho-Kashyap procedure. The proposed method uses the absolute error, rather than the squared error, to design a linear classifier. Additionally, easy control of the generalization ability and robustness to outliers are obtained. Next, an extension to a nonlinear classifier by the mixture-of-experts technique is presented. Each expert is represented by a fuzzy if-then rule in the Takagi-Sugeno-Kang form. Finally, examples are given to demonstrate the validity of the introduced method.
The fuzzy c-means method is one of the most popular clustering methods based on minimization of a criterion function. However, one of the greatest disadvantages of this method is its sensitivity to the presence of noise and outliers in data. The epsilon -insensitive Fuzzy C-Means ( epsilon FCM) clustering algorithm is free of this disadvantage, but has a very high computational burden and requires a choice of the insensitivity parameter(s) epsilon In this paper, a new computationally effective epsilon -insensitive fuzzy c-means clustering algorithm with automatic adjustment of the insensitivity parameter(s) is introduced. Performance of the new clustering algorithm is experimentally verified using synthetic data with outliers and overlapped groups of heavy-tailed data.
16
Dostęp do pełnego tekstu na zewnętrznej witrynie WWW
The Hard C-Means (HCM) clustering method is one of the most popular clustering methods based on minimization of a criterion function. However, one of the greates disadvantage of this method is that the performance of the HCM is good only when the data set contains clusters that have approximately the same size and shape. The paper is devoted to a new clustering algorithm, called minimum hypervolume clustering (MHC), that seeks C hyperellipsoids with the smallest hypervolumes that enclose all the data points. Performances of the new clustering algorithm are experimentally verified using synthetic and real life data containing clusters with different size and orientations.
17
Dostęp do pełnego tekstu na zewnętrznej witrynie WWW
A new learning method tolerant of imprecision is introduced and used in neuro-fuzzy modelling. The proposed method makes it possible to dispose of an intrinsic inconsistency of neuro-fuzzy modelling, where zero-tolerance learning is used to obtain a fuzzy model tolerant of imprecision. This new method can be called e-insensitive learning, where, in order to fit the fuzzy model to real data, the e-insensitive loss function is used. e-insensitive learning leads to a model with minimal Vapnik-Chervonenkis dimension, which results in an improved generalization ability of this system. Another advantage of the proposed method is its robustness against outliers. This paper introduces two approaches to solving e-insensitive learning problem. The first approach leads to a quadratic programming problem with bound constraints and one linear equality constraint. The second approach leads to a problem of solving a system of linear inequalities. Two computationally efficient numerical methods for e-insensitive learning are proposed. Finally, examples are given to demonstrate the validity of the introduced methods.
18
Dostęp do pełnego tekstu na zewnętrznej witrynie WWW
Fuzzy clustering helps to find natural vague boundaries in data. The fuzzy C-Means (FCM) method is one of the most popular clustering methods based on minimization of a criterion function. However, one of the greatest disadvantages of this method is its sensivity to presence of noise and outliers in data. This paper introduces a new ε-insensitive Fuzzy C-Medians (εFCMed) clustering algorithm. As a special case, this algorithm includes the well-known Fuzzy C-Medians method (FCMed). Performance of the new clustering algorithm is experimentally compared with the Fuzzy C-Means (FCM) and the FCMed methods using synthetic heavy-tailed and overlapped groups in background noise, and the Iris database.
19
Dostęp do pełnego tekstu na zewnętrznej witrynie WWW
This paper introduces a new classifier design method, that is based on an extension of the classical Ho-Kashyap procedure. The proposed method uses absolute error rather than square errorto design a linear classifier. Additionally, easy control of generalization ability and outliers robustness is obtained. Finally, examples are giver to demonstrate the validity of the introduced method.
20
Dostęp do pełnego tekstu na zewnętrznej witrynie WWW
Fuzzy clustering can be helpful in finding natural vague boundaries in data. The fuzzy c-means method is one of the most popular clustering methods based on minimization of a criterion function. However, one of the greatest disadvantages of this method is its sensitivity to the presence of noise and outliers in the data. The present paper introduces a new varepsilon-insensitive Fuzzy C-Means (varepsilonFCM) clustering algorithm. As a special case, this algorithm includes the well-known Fuzzy C-Medians method (FCMED). The performance of the new clustering algorithm is experimentally compared with the Fuzzy C-Means (FCM) method using synthetic data with outliers and heavy-tailed, overlapped groups of the data.
JavaScript jest wyłączony w Twojej przeglądarce internetowej. Włącz go, a następnie odśwież stronę, aby móc w pełni z niej korzystać.