Wyniki wyszukiwania - BazTech

1

An intelligent neural network algorithm for uncertainty handling in sensor failure scenario of food quality assurance model

Deepa S. N., Jayalakshmi N. Yogamba

Computer Assisted Methods in Engineering and Science

|

2022

|

Vol. 29, no. 1-2 spec.

105--123

EN

The quality of food is usually tested by sensing the product odor using e-nose technique.However, in a real-time testing environment, some of the employed sensors may fail tooperate, which imposes great uncertainty on the food quality assurance model. To handlethe uncertainty, a support vector machine (SVM) classifier algorithm is developed todeal with the failure sensor effect using a data imputation strategy. The proposed modelis evaluated experimentally by means of benchmark datasets, and validated in a real-time environment by programming an Arduino-UNO controller in the internet of things(IoT) environment.

2

An automated classification of HEp-2 cellular shapes using Bag-of-keypoint features and Ant Colony Optimization

Devanathan Kanchana, Ganesan Kavitha, Swaminathan Ramakrishnan

Biocybernetics and Biomedical Engineering

|

2021

|

Vol. 41, no. 2

376--390

EN

In this study, an attempt has been made to differentiate HEp-2 cellular shapes using Bag-of keypoint features and optimization. For this, the images are considered from a publicly available database. To increase the cell structure visibility, the images are pre-processed using edge-sensitive local contrast enhancement. Further, the Speeded-up Robust Feature (SURF) keypoints are extracted and Bag-of-keypoints for each shape are generated. These features are subjected to Ant Colony Optimization (ACO) algorithm for feature selection. The optimal features obtained are then fed to Support Vector Machine (SVM) and k-Nearest Neighbour (kNN) classifiers. Results show that the ACO algorithm can identify the optimal features that characterize the cellular shapes. SVM and kNN are able to differentiate between the shapes with an average classification accuracy of 93.6% and 94.8% respectively. Since differential diagnosis of HEp-2 cellular shapes is significant in the disease-specific prognosis and treatment, this study seems to be clinically relevant.

3

Curve skeleton extraction via k-nearest-neighbors based contraction

Zhou Jianling, Liu Ji, Zhang Min

International Journal of Applied Mathematics and Computer Science

|

2020

|

Vol. 30, no. 1

123--132

EN

We propose a skeletonization algorithm that is based on an iterative points contraction. We make an observation that the local center that is obtained via optimizing the sum of the distance to k nearest neighbors possesses good properties of robustness to noise and incomplete data. Based on such an observation, we devise a skeletonization algorithm that mainly consists of two stages: points contraction and skeleton nodes connection. Extensive experiments show that our method can work on raw scans of real-world objects and exhibits better robustness than the previous results in terms of extracting topology-preserving curve skeletons.

4

Complex-valued distribution entropy and its application for seizure detection

Zhang Tao, Chen Wanzhong, Li Mingyang

Biocybernetics and Biomedical Engineering

|

2020

|

Vol. 40, no. 1

306--323

EN

Embedding entropies are powerful indicators in quantifying the complexity of signal, but most of them are only applicable for real-valued signal and the phase information is ignored if the analyzed signal is complex-valued. To assess the complexity of complex-valued signal, a new entropy called complex-valued distribution entropy (CVDistEn) was first proposed in this study. Two rules, namely equal width criterion and equal area criterion, were employed to demarcate the complex-valued space and two kinds of CVDistEn, i.e., CVDistEn1 and CVDistEn2 were raised. Furthermore, two novel feature extraction methods: (1) flexible analytic wavelet transform (FAWT)-based CVDistEn1 and logarithmic energy (LE) (FAWTC1L), (2) FAWT-based CVDistEn2 and LE (FAWTC2L) were subsequently put forward to characterize the interictal and ictal EEGs. Fuzzy k-nearest neighbors (FKNN) classifier was finally employed to classify these two types of EEGs automatically. Experiment results show the fusion method of FAWTC1L and FKNN leads to the best accuracies (ACCs)/Matthews correlation coefficients (MCCs) of 99.99%/99.97% and 100%/100% for Bonn and Neurology & Sleep Centre EEG datasets, respectively, while the other fusion scheme of FAWTC2L and FKNN results in the highest ACCs/MCCs of 99.97%/99.93% and 99.94%/99.89% for the same datasets. The proposed methods outperform other entropy-related seizure detection schemes and most of state-of-the-art techniques, they provide another new way for automated seizure detection in EEG.

5

Comparison of prototype selection algorithms used in construction of neural networks learned by SVD

Jankowski N.

International Journal of Applied Mathematics and Computer Science

|

2018

|

Vol. 28, no. 4

719--733

EN

Radial basis function networks (RBFNs) or extreme learning machines (ELMs) can be seen as linear combinations of kernel functions (hidden neurons). Kernels can be constructed in random processes like in ELMs, or the positions of kernels can be initialized by a random subset of training vectors, or kernels can be constructed in a (sub-)learning process (sometimes by k-means, for example). We found that kernels constructed using prototype selection algorithms provide very accurate and stable solutions. What is more, prototype selection algorithms automatically choose not only the placement of prototypes, but also their number. Thanks to this advantage, it is no longer necessary to estimate the number of kernels with time-consuming multiple train-test procedures. The best results of learning can be obtained by pseudo-inverse learning with a singular value decomposition (SVD) algorithm. The article presents a comparison of several prototype selection algorithms co-working with singular value decomposition-based learning. The presented comparison clearly shows that the combination of prototype selection and SVD learning of a neural network is significantly better than a random selection of kernels for the RBFN or the ELM, the support vector machine or the kNN. Moreover, the presented learning scheme requires no parameters except for the width of the Gaussian kernel.

6

Detection of valvular heart diseases using impedance cardiography ICG

Chabchoub S., Mansouri S., Ben Salah R.

Biocybernetics and Biomedical Engineering

|

2018

|

Vol. 38, no. 2

251--261

EN

Impedance cardiography (ICG) is a simple, non-invasive and cost effective tool for monitor-ing hemodynamic parameters. It has been successfully used to diagnose several cardiovas-cular diseases, like the heart failure and myocardial infarction. In particular, valvular heart disease (VHD) is characterized by the affection of one or more heart valves: mitral, aortic, tricuspid or pulmonary valves and it is usually diagnosed using the Doppler echocardiogra- phy. However, this technique is rather expensive, requires qualified expertise, discontinu- ous, and often not necessary to make just a simple diagnosis. In this paper, a new computer aided diagnosis system is proposed to detect VHD using the ICG signals. Six types of ICG heartbeats are analyzed and classified: normal heartbeats (N), mitral insufficiency heart-beats (MI), aortic insufficiency heartbeats (AI), mitral stenosis heartbeats (MS), aortic steno-sis heartbeats (AS), and pulmonary stenosis heartbeats (PS). The proposed methodology is validated on 120 ICG recordings. Firstly, ICG signal is denoised using the Daubechies wavelet family with order eight (db8). Then, these signals are segmented into several heartbeats and, later, subjected to the linear prediction LP and discrete wavelet transform DWT approaches to extract temporal and time–frequency features, respectively. In order to reduce the number of features and select the most relevant ones among them, the Student's t-test is applied. Therefore, a total of 16 features are selected (3 temporal features and 13 time– frequency features). For the classification step, the support vector machine SVM and k-nearest neighbors KNN classifiers are used. Different combinations between extracted features and classifiers are proposed. Hence, experimental results showed that the combi-nation between temporal features, time–frequency features and SVM classifier achieved the highest classification performance in classifying the N, MI, MS, AI, AS and PS heartbeats with 98.94% of overall accuracy.

7

A bionic hand controlled by hand gesture recognition based on surface EMG signals: A preliminary study

Shi W. T., Lyu Z. J., Tang S. T., Chia T. L., Yang C. Y.

Biocybernetics and Biomedical Engineering

|

2018

|

Vol. 38, no. 1

126--135

EN

A bionic hand with fine motor ability could be a favorable option for replacing the human hand when performing various operations. Myoelectric control has been widely used to recognize hand movements in recent years. However, most of the previous studies have focused on whole-hand movements, with only a few investigating subtler motions. The aim of this study was to construct a prototype system for recognizing hand postures with the aim of controlling a bionic hand by analyzing sEMG signals measured at the flexor digitorum superficialis and extensor digitorum muscles. We adopted multiple features commonly used in previous studies—mean absolute value, zero crossing, slope sign change, and waveform length—in the algorithm for extracting hand-posture features, and the k-nearest-neighbors (KNN) algorithm as the classifier to perform hand-posture recognition. The bionic hand was controlled by an Arduino microprocessor, which converted the signals received from the classification process that were fed to the servo motors controlling the bionic fingers. We constructed a two-channel sEMG pattern-recognition system that can identify human hand postures and control a homemade bionic hand to perform corresponding hand postures. The KNN approach was able to recognize four different hand postures with a classification accuracy of 94% in the online experiment by using the channel combination. Moreover, the experimental tests show that the bionic hand could faithfully imitate the hand postures of the human hand. This study has bridged the gap between the features of sEMG signals of fingers and the postures of a bionic hand.

8

Prognozowanie miesięcznego zapotrzebowania na energię elektryczną metodą k najbliższych sąsiadów

Dudek G., Pełka P.

Przegląd Elektrotechniczny

|

2017

|

R. 93, nr 4

62--65

PL

Przedstawiono model prognostyczny oparty na metodzie k najbliższych sąsiadów do prognozowania miesięcznego zapotrzebowania na energię elektryczną. Model wykorzystuje analogie pomiędzy fragmentami szeregów czasowych reprezentowanymi przez ich obrazy. Obrazy zapewniają ujednolicenie danych wejściowych i wyjściowych, odfiltrowanie trendu i uproszczenie modelowanej zależności. W części eksperymentalnej model przetestowano w prognozach dla wybranych państw europejskich.

EN

A forecasting model based on the k nearest neighbor method for forecasting monthly electricity demand is presented. The model uses analogies between fragments of time series represented by their patterns. Patterns ensure unification of input and output data, filtering out the trend and simplification of the modeled relationship. In the experimental part of the work the model was tested in forecasting for selected European countries.

9

Application of the k nearest neighbors method to fuzzy data processing

Pluciński M., Pietrzykowski M.

Przegląd Elektrotechniczny

|

2017

|

R. 93, nr 1

77--81

EN

The paper presents that with the application of fuzzy numbers arithmetic, the k nearest neighbors method can be adapted to various types of data. Both, the learning data and the input data may be in the form of the crisp number, interval or fuzzy number. Experiments proved that the method works correctly and gives credible results. There is also shown that the kNN method can be used for the determination of the fuzzy model output.

PL

W artykule pokazano, że z wykorzystaniem arytmetyki rozmytej, metoda k najbliższych sąsiadów może być zastosowana do danych różnego typu. Zarówno dane uczące, jak i dane wejściowe modelu mogą być liczbami, interwałami lub liczbami rozmytymi. Eksperymenty wykazały, że metoda działa prwidłowo i daje wiarygodne wyniki. Zaprezentowano również możliwość użycia metody k najbliższych sąsiadów do wyznaczania wyjścia modelu rozmytego.

10

Classification of falling asleep states using HRV analysis

Piotrowski Z., Szypulska M.

Biocybernetics and Biomedical Engineering

|

2017

|

Vol. 37, no. 2

290--301

EN

The article presents the results of studies on drowsiness and drowsiness detection performed using heart rate variability analysis (HRV). The results of those studies indicate that the most significant parameters, from the standpoint of classification of drowsiness are the following parameters of the HRV analysis: the low and high frequency band the ratio of the tachogram power in the LF and HF bands, and the total power distribution. The best detection results were obtained for the following methods, in the following order: the nearest neighborhood with metrics: standardized Euclides and Mahalanobis, the square discriminant analysis, and the Bayesian classifier. In order to classify drowsiness periods, a neural network was also used; it consisted of four inputs, six neurons in the hidden layer, and three outputs, one of which was assigned to one of the accepted classes. In order to obtain the most effective learning, a linear feed forward network was designed using back propagation of errors and the RPROP algorithm. In the case of this type of networks, the achieved accuracy of the individual classes was on the level of 98.7%.

11

Predykcja poziomu tlenu w piecu EAF z wykorzystaniem metod sztucznej inteligencji : regresji liniowej i metody najblizszych sąsiadów (k-NN)

Drzewiecki P.

Piece Przemysłowe & Kotły

|

2013

|

1-2

30--37

PL

W obecnych czasach ludzki intelekt zaczyna zastępować sztuczna inteligencja i związane z nią metody predykcji, czyli przewidywania, na podstawie zgromadzonych wcześniej danych. Nie ma firmy, która by nie starała się przewidzieć swoich potencjalnych zysków np na podstawie różnych danych statystycznych ustalając jednocześnie, który czynnik ma największy wpływ na rozwój firmy. Z tych przyczyn również sztuczna inteligencja znalazła zastosowanie w różnych dziedzinach przemysłu ,gdzie nawet najdrobniejsza poprawa parametrów procesu może powodować znaczne obniżenie kosztów produkcji i wpłynąć bezpośrednio na wzrost konkurencyjności firmy na rynku. Niniejsze opracowanie ma na celu przedstawienie metod sztucznej inteligencje pozwalające na analizę wpływu parametrów procesu stalowniczego na stopień aktywności tlenu, przy użyciu regresji linowej, nieliniowej i nieparametrycznej a następnie na wyborze najlepszego modelu a następnie porównanie ich.

12

Categorization of Similar Objects Using Bag of Visual Words and k - Nearest Neighbour Classifier

Artiemjew P., Górecki P., Sopyła K.

Technical Sciences / University of Warmia and Mazury in Olsztyn

|

2012

|

nr 15(2)

293-305

EN

Image categorization is one of the fundamental tasks in computer vision, it has wide application in methods of artificial intelligence, robotic vision and many others. There are a lot of difficulties in computer vision to overcome, one of them appears during image recognition and classification. The difficulty arises from an image variance, which may be caused by scaling, rotation, changes in a perspective, illumination levels, or partial occlusions. Due to these reasons, the main task is to represent represent images in such way that would allow recognizing them even if they have been modified. Bag of Visual Words (BoVW) approach, which allows for describing local characteristic features of images, has recently gained much attention in the computer vision community. In this article we have presented the results of image classification with the use of BoVW and k - Nearest Neighbor classifier with different kinds of metrics and similarity measures. Additionally, the results of k - NN classification are compared with the ones obtained from a Support Vector Machine classifier.

13

Nonlinear ellipsoidal mini-models – application for the function approximation task

Pluciński M.

Przegląd Elektrotechniczny

|

2012

|

R. 88, nr 10b

247-251

EN

Mini-models are local regression models, which can be used for the function approximation learning. In the paper, there are presented mini-models based on hyper-spheres and hyper-ellipsoids and researches were made for linear and nonlinear models with no limitations for the problem input space dimension. Learning of the approximation function based on mini-models is very fast and it proved to have a good accuracy. Mini-models have also very advantageous extrapolation properties.

PL

Mini-modele to modele lokalnej regresji, które można wykorzystać do aproksymacji funkcji. W artykule opisano mini-modele o bazie hiper-sferycznej i hiper-elipsoidalnej oraz badania dla mini-modeli linowych i nieliniowych bez ograniczeń na rozmiar przestrzeni wejść. Uczenie aproksymującej funkcji opartej na mini-modelach jest szybkie, a sama funkcja ma dobrą dokładność i korzystne własności ekstrapolacyjne.

14

Application of the information-gap theory for evaluation of nearest neighbours method robustness to data uncertainty

Pluciński M.

Przegląd Elektrotechniczny

|

2012

|

R. 88, nr 10b

272-275

EN

The paper describes a new method based on the information-gap theory which enables an evaluation of worst case error predictions of the kNN method in the presence of a specified level of uncertainty in the data. There are presented concepts of a robustness and an opportunity of the kNN model and calculations of these concepts were performed for a simple 1-D data set and next, for a more complicated 6-D data set. In both cases the method worked correctly and enabled evaluation of the robustness and the opportunity for a given lowest acceptable quality rc or a windfall quality rw. The method enabled also choosing of the most robust kNN model for a given level of an uncertainty [alfa].

PL

W artykule opisane jest zastosowanie teorii luk informacyjnych do określania największego błędu modelu kNN w przypadku wystąpienia w danych niepewności o określonym poziomie. Przedstawione zostały pojęcia odporności i sposobności modelu kNN oraz pokazane zostały przykłady ich wyznaczania dla prostych danych jednowejściowych i bardziej złożonych, sześciowejściowych. W obu przypadkach metoda działała prawidłowo, a dodatkowo umożliwiała wyznaczanie najbardziej odpornego modelu kNN przy określonym poziomie niepewności [alfa].

15

Przetwarzanie danych w opartych na podobieństwie metodach prognozowania przebiegów dobowych zapotrzebowania na moc elektryczną

Dudek G.

Przegląd Elektrotechniczny

|

2006

|

R. 82, nr 9

15-19

PL

Przedstawiono koncepcję metod prognozowania wykorzystujących podobieństwo obrazów sekwencji szeregów czasowych. Podano kilka metod przetwarzania elementów sekwencji szeregu czasowego w celu konstrukcji obrazów wejściowych i obrazów prognoz. Dla różnych definicji obrazów wykonano prognozy dobowych przebiegów obciążeń systemu elektroenergetycznego używając minimalnoodległościowego modelu prognostycznego k najbliższych sąsiadów.

EN

A conception of forecasting methods based on similarities between time series sequence patterns is presented. Several methods of time series sequence elements preprocessing for forming input patterns and forecast patterns are described. For different definitions of patterns, forecasts of the daily power system load profiles are constructed. For this purpose a minimum-distance method, based on the k-nearest-neighbor rule, was applied.