Wyniki wyszukiwania - BazTech

1

Classification of cognitive states using clustering-split time series framework

Ramakrishna J. Siva, Ramasangu Hariharan

Computer Assisted Methods in Engineering and Science

|

2024

|

Vol. 31, no. 2

241--260

EN

Over the last two decades, functional Magnetic Resonance Imaging (fMRI) has provided immense data about the dynamics of the brain. Ongoing developments in machine learning suggest improvements in the performance of fMRI data analysis. Clustering is one of the critical techniques in machine learning. Unsupervised clustering techniques are utilized to partition the data objects into different groups. Supervised classification techniques applied to fMRI data facilitate the decoding of cognitive states while a subject is engaged in a cognitive task. Due to the high dimensional, sparse, and noisy nature of fMRI data, designing a classifier model for estimating cognitive states becomes challenging. Feature selection and feature extraction techniques are critical aspects of fMRI data analysis. In this work, we present one such synergy, a combination of Hierarchical Consensus Clustering (HCC) and the Statistics of Split Timeseries (SST) framework to estimate cognitive states. The proposed HCC-SST model’s performance has been verified on StarPlus fMRI data. The obtained experimental results show that the proposed classifier model achieves 99% classification accuracy with a smaller number of voxels and lower computational cost.

2

Analysis of the possibility of mechanical devices supervision based on the measurement of the vibration level and SVM classifiers

Schossler Ryszard, Grochowina Marcin, Leniowska Lucyna

Vibrations in Physical Systems

|

2022

|

Vol. 33, nr 3

art. no. 2022316

EN

This paper presents the concept of diagnosing the technical condition of mechanical devices. The test is based on a non-invasive vibration analysis technique combined with the use of artificial intelligence method. The object of the research is an electric motor for which vibrations were recorded by a vibration sensor based on four 3-axis digital accelerometers and MPU-6050 gyroscopes. The effectiveness of classification methods using the two-class and one-class classification was compared. It has been shown that the use of an incomplete pattern of the vibration model and a single-class classifier allows for effective detection of anomalies in the operation of an induction motor. Satisfactory classification efficiency was achieved, despite the limitation of the teaching set only to the information obtained during the correct operation of the device. The described method is universal and can be used to diagnose the technical condition of many different types of technical devices.

3

Klasyfikator SVM w zastosowaniu do synchronizacji sygnału OFDM zniekształconego przez kanał wielodrogowy

Blok Marek, Gmiński Karol, Miklaszewski Patryk

Przegląd Telekomunikacyjny + Wiadomości Telekomunikacyjne

|

2022

|

nr 4

412--415

PL

W pracy przedstawiono analizę przydatności klasyfikatora SVM bazującego na uczeniu maszynowym do estymacji przesunięcia czasowego odebranego symbolu OFDM. Przedstawione wyniki wykazują, że ten klasyfikator potrafi zapewnić synchronizację dla różnych kanałów wielodrogowych o wysokim poziomie szumu. Eksperymenty przeprowadzone w Matlabie z użyciem modeli modulatora i demodulatora wykazały, że w większości przypadków klasyfikator prawidłowo oceniał przesunięcie czasowe symbolu OFDM, nawet gdy był trenowany dla innego modelu kanału.

EN

The paper presents an analysis of capabilities of machine learning based SVM classifier to estimate the OFDM symbol time offset at the receiver. The presented results demonstrate classifier’s ability to provide synchronization for various multipath channel models with high noise levels. The research was conducted using OFDM modulator and demodulator models implemented in Matlab. It has been shown that in most cases the classifier was able to determine the time shift of the OFDM symbol with high confidence even if it was trained for a different channel model.

4

Human Face Expressions from Images

Pilarczyk Rafał, Chang Xin, Skarbek Władysław

Fundamenta Informaticae

|

2019

|

Vol. 168, nr 2-4

287--310

EN

Several computer algorithms for recognition of visible human emotions are compared at the web camera scenario using CNN/MMOD face detector. The recognition refers to four face expressions: smile, surprise, anger, and neutral. At the feature extraction stage, the following three concepts of face description are confronted: (a) static 2D face geometry represented by its 68 characteristic landmarks (FP68); (b) dynamic 3D geometry defined by motion parameters for eight distinguished face parts (denoted as AU8) of personalized Candide-3 model; (c) static 2D visual description as 2D array of gray scale pixels (known as facial raw image). At the classification stage, the performance of two major models are analyzed: (a) support vector machine (SVM) with kernel options; (b) convolutional neural network (CNN) with variety of relevant tensor processing layers and blocks of them. The models are trained for frontal views of human faces while they are tested for arbitrary head poses. For geometric features, the success rate (accuracy) indicate nearly triple increase of performance of CNN with respect to SVM classifiers. For raw images, CNN outperforms in accuracy its best geometric counterpart (AU/CNN) by about 30 percent while the best SVM solutions are inferior. For F-score the high advantage of raw/CNN over geometric/CNN and geometric/SVM is observed, as well. We conclude that contrary to CNN based emotion classifiers, the generalization capability wrt human head pose for SVM based emotion classifiers, is worse too.

5

Machine learning methods for MRI biomarkers analysis of pediatric posterior fossa tumors

Li Mengmeng, Shang Zhigang, Yang Zhongliang, Zhang Yong, Wan Hong

Biocybernetics and Biomedical Engineering

|

2019

|

Vol. 39, no. 3

765--774

EN

Medical imaging technologies provide an increasing number of opportunities for disease prediction and prognosis. Specifically, imaging biomarkers can quantify the entire tumor phenotypes to enhance the prediction. Machine learning technology can be explored to mine and analyze these biomarkers and to establish predictive models for the clinical applications. Several studies have applied various machine learning methods to imaging biomark-ers based clinical predictions of different diseases. Here we seek to evaluate different machine learning methods in pediatric posterior fossa tumor prediction. We present a machine learning based magnetic resonance imaging biomarkers analysis framework for two kinds of pediatric posterior fossa tumors. In details, three feature extraction methods are used to obtain 300 imaging biomarkers. 10 feature selection methods and 11 classifiers are evaluated by the quantified predictive performance and stability, and importance consistency of features and the influence of the experimental factors are also analyzed. Our results demonstrate that the CFS feature selection method (accuracy: 83.85 5.51%, stability: [0.84, 0.06]) and SVM classifier (accuracy: 85.38 3.47%, RSD: 4.77%) show relatively better performance than others and should be preferred. Among all the biomarkers, 17 texture features seem to be more important. Multifactor analysis results indicate the choice of classifier accounts for the most contribution to the variability in performance (37.25%). The machine learning based framework is efficient for pediatric posterior fossa tumors biomarkers analysis and could provide valuable references and decision support for assisted clinical diagnosis.

6

Automated quantification of ultrasonic fatty liver texture based on curvelet transform and SVD

Bharath R., Mishra P. K., Rajalakshmi P.

Biocybernetics and Biomedical Engineering

|

2018

|

Vol. 38, no. 1

145--157

EN

Fatty liver is a prevalent disease and is the major cause for the dysfunction of the liver. If fatty liver is untreated, it may progress into chronic diseases like cirrhosis, hepatocellular carcinoma, liver cancer, etc. Early and accurate detection of fatty liver is crucial to prevent the fatty liver progressing into chronic diseases. Based on the severity of fat, the liver is categorized into four classes, namely Normal, Grade I, Grade II and Grade III respectively. Ultrasound scanning is the widely used imaging modality for diagnosing the fatty liver. The ultrasonic texture of liver parenchyma is specific to the severity of fat present in the liver and hence we formulated the quantification of fatty liver as a texture discrimination problem. In this paper, we propose a novel algorithm to discriminate the texture of fatty liver based on curvelet transform and SVD. Initially, the texture image is decomposed into sub-band images with curvelet transform enhancing gradients and curves in the texture, then an absolute mean of the singular values are extracted from each curvelet decomposed image, and used it as a feature representation for the texture. Finally, a cubic SVM classifier is used to classify the texture based on the extracted features. Tested on a database of 1000 image textures with 250 image textures belonging to each class, the proposed algorithm gave an accuracy of 96.9% in classifying the four grades of fat in the liver.

7

Dobór cech diagnostycznych dla klasyfikatora SVM w zadaniu klasyfikacji stanu przetoki tętniczo-żylnej na podstawie sygnału akustycznego

Grochowina M., Leniowska L.

Acta Bio-Optica et Informatica Medica. Inżynieria Biomedyczna

|

2016

|

Vol. 22, nr 4

207--212

PL

W artykule przedstawiono proces selekcji cech diagnostycznych dla klasyfikatora SVM. Badania przeprowadzone zostały z użyciem zbioru danych zawierającego próbki sygnału dźwiękowego emitowanego przez przetokę tętniczo-żylną. Celem prac było stworzenie rozwiązania klasyfikacji wieloklasowej w oparciu o klasyfikator z rodziny SVM pozwalającego na skuteczną i wiarygodną ocenę stanu przetoki tętniczo-żylnej.

EN

The paper presents the process of selection diagnostic features for SVM classifier. The study was conducted with using a data set containing samples of the sound signal emitted by the arteriovenous fistula. The objective was to create a solution multi-class classification based on SVM classifier family allowing for an effective and reliable evaluation of the arteriovenous fistula state.

8

Genetic algorithm and forward method for feature selection in EEG feature space

Rejer I., Lorenz K.

Journal of Theoretical and Applied Computer Science

|

2013

|

Vol. 7, nr 2

72--82

EN

There are a lot of problems that arise in the process of building a brain-computer interface based on electroencephalographic signals (EEG). A huge imbalance between a number of experiments possible to conduct and the size of feature space, containing features extracted from recorded signals, is one of them. To reduce this imbalance, it is necessary to apply methods for feature selection. One of the approaches for feature selection, often taken in brain-computer interface researches, is a classic genetic algorithm that codes all features within each individual. In this study, there will be shown, that although this approach allows obtaining a set of features of high classification precision, it also leads to a feature set highly redundant comparing to a set of features selected using a forward selection method or a genetic algorithm equipped with individuals of a given (very small) number of genes.

9

Classification of animals to determine the migration potential at the construction of new infrastructure

Matuska S., Hudec R., Benco M., Zachariasova M.

Archives of Transport System Telematics

|

2013

|

Vol. 6, iss. 4

26--30

EN

At the planning and construction of new infrastructures, the information about migration potential of animals in a target area is needed. This information will be used to design of migration corridors for wild animals. To determine the migration potential of animals based on distributed video camera system, new methods for object recognition and classification are developed. In general, an object recognition system consists of three steps, namely, the image feature extraction from the training database, training the classifier and evaluation of query image of object/animal. In this paper, an extraction of local key point by SIFT or SURF descriptors, bags of key points method in combination with SVM classifier and two hybrid key points detection methods are proposed in detail.

10

Investigation of normalization techniques and their impact on a recognition rate in handwritten numeral recognition

Chmielnicki W., Stąpor K.

Schedae Informaticae

|

2010

|

Vol. 19

53-77

EN

This paper presents several normalization techniques used in handwritten numeral recognition and their impact on recognition rates. Experiments with five different feature vectors based on geometric invariants, Zernike moments and gradient features are conducted. The recognition rates obtained using combination of these methods with gradient features and the SVM-rbf classifier are comparable to the best state-of-art techniques.