Wyniki wyszukiwania - BazTech

1

Semi-supervised vs. supervised learning for mental health monitoring: A case study on bipolar disorder

Casalino Gabriella, Castellano Giovanna, Hryniewicz Olgierd, Leite Daniel, Opara Karol, Radziszewska Weronika, Kaczmarek-Majer Katarzyna

International Journal of Applied Mathematics and Computer Science

|

2023

|

Vol. 33, no. 3

419--428

EN

Acoustic features of speech are promising as objective markers for mental health monitoring. Specialized smartphone apps can gather such acoustic data without disrupting the daily activities of patients. Nonetheless, the psychiatric assessment of the patient’s mental state is typically a sporadic occurrence that takes place every few months. Consequently, only a slight fraction of the acoustic data is labeled and applicable for supervised learning. The majority of the related work on mental health monitoring limits the considerations only to labeled data using a predefined ground-truth period. On the other hand, semi-supervised methods make it possible to utilize the entire dataset, exploiting the regularities in the unlabeled portion of the data to improve the predictive power of a model. To assess the applicability of semi-supervised learning approaches, we discuss selected state-of-the-art semi-supervised classifiers, namely, label spreading, label propagation, a semi-supervised support vector machine, and the self training classifier. We use real-world data obtained from a bipolar disorder patient to compare the performance of the different methods with that of baseline supervised learning methods. The experiment shows that semi-supervised learning algorithms can outperform supervised algorithms in predicting bipolar disorder episodes.

2

Diagnosing faults in the timing system of a passenger car spark ignition engine using the bayes classifier and entropy of vibration signals

Czech Piotr

Zeszyty Naukowe. Transport / Politechnika Śląska

|

2022

|

z. 116

83--98

EN

Today's systems for diagnosing the technical condition of machines, including vehicles, use very advanced methods of acquiring and processing input data. Presently, work is being conducted globally to solve related problems. At the moment, it is not yet possible to create a single procedure that would enable the construction of a properly functioning diagnostic system, regardless of the selected object to be diagnosed. Hence, there is a need to conduct further research into the possibility of using already developed methods, as well as their modification to other diagnostic cases. This article presents the results of research related to the use of the Bayes classifier for diagnosing the technical condition of passenger car engine components. Damage to the exhaust valve of a spark ignition engine was diagnosed. The source of information on the technical condition was vibration signals recorded at various measuring points and under different operating conditions of the car. To describe the nature of changes in the vibration signals, the entropy measures were determined for the decomposed signal using the discrete wavelet transform is proposed.

3

Classification of epileptic EEG signals using PSO based artificial neural network and tunable-Q wavelet transform

George S. Thomas, Subathra M. S. P., Sairamya N. J., Susmitha L., Joel Premkumar M.

Biocybernetics and Biomedical Engineering

|

2020

|

Vol. 40, no. 2

709--728

EN

Epilepsy is a widely spread neurological disorder caused due to the abnormal excessive neural activity which can be diagnosed by inspecting the electroencephalography (EEG) signals visually. The manual inspection of EEG signals is subjected to human error and is a tedious process. Further, an accurate diagnosis of generalized and focal epileptic seizures from normal EEG signals is vital for the supervision of pertinent treatment, life advancement of the subjects, and reduction in cost for the subjects. Hence the development of automatic detection of generalized and focal epileptic seizures from normal EEG signals is important. An approach based on tunable-Q wavelet transform (TQWT), entropies, Particle Swarm Optimization (PSO) and Artificial Neural Network (ANN) is proposed in this work for detection of epileptic seizures and its types. Two EEG databases namely, Karunya Institute of Technology and Sciences (KITS) EEG database and Temple University Hospital (TUH) database consisting of normal, generalized and focal EEG signals is used in this work to analyze the performance of the proposed approach. Initially, the EEG signals are decomposed into sub-bands using TQWT and the non-linear features like log energy entropy, Shannon entropy and Stein's unbiased risk estimate (SURE) entropy is computed from each sub-band. The informative features from the computed feature vectors are selected using PSO and fed into ANN for the classification of EEG signals. The proposed algorithm for KITS database achieved a maximum accuracy of 100% for four experimental cases namely, (i) normal-focal, (ii) normal-generalised, (iii) normal-focal + generalised and (iv) normal-focal-generalised. The TUH database achieved an accuracy of 95.1%, 97.4%, 96.2% and 88.8% for the four experimental cases. The proposed approach is promising and able to discriminate the epileptic seizure types with satisfactory classification performance.

4

Single channel EMG-based continuous terrain identification with simple classifier for lower limb prosthesis

Gupta Rohit, Agarwal Ravinder

Biocybernetics and Biomedical Engineering

|

2019

|

Vol. 39, no. 3

775--788

EN

The focus of the present research endeavour is to propose a single channel Electromyogram (EMG) signal driven continuous terrain identification method utilizing a simple classifier. An iterative feature selection algorithm has also been proposed to provide effective information to the classifiers. The proposed method has been validated on EMG signal of fifteen subjects and ten subjects for three and five daily life terrains respectively. Feature selection algorithm has significantly improved the identification accuracy (ANOVA, p-value < 0.05) as compared to principal component analysis (PCA) technique. The average identification accuracies obtained by Support Vector Machine (SVM), Linear Discriminant Analysis (LDA) and Neural Network (NN) classifiers are 96.83 ± 0.28%, 97.45 ± 0.32% and 97.61 ± 0.22% respectively. Subject wise performance (five subjects) of individually trained classifiers shows no significant degradation and difference in performance among the subjects even for the untrained data (ANOVA, p-value > 0.05). The study has been extended to dual muscle approach for terrain identification. However, the proposed algorithm has shown similar performance even with the single muscle approach (ANOVA, p-value > 0.05). The outcome of the proposed continuous terrain identifi-cation method shows a pronounced potential in efficient lower limb prosthesis control.

5

Closest paths in graph drawings under an elastic metric

Baran M.

International Journal of Applied Mathematics and Computer Science

|

2018

|

Vol. 28, no. 2

387--397

EN

This work extends the dynamic programming approach to calculation of an elastic metric between two curves to finding paths in pairs of graph drawings that are closest under this metric. The new algorithm effectively solves this problem when all paths between two given nodes in one of these graphs have the same length. It is then applied to the problem of pattern recognition constrained by a superpixel segmentation. Segmentations of test images, obtained without statistical modeling given two shape endpoints, have good accuracy.

6

A connectionist computational method for face recognition

Pujol F. A., Mora H., Girona-Selva J. A.

International Journal of Applied Mathematics and Computer Science

|

2016

|

Vol. 26, no. 2

451--465

EN

In this work, a modified version of the elastic bunch graph matching (EBGM) algorithm for face recognition is introduced. First, faces are detected by using a fuzzy skin detector based on the RGB color space. Then, the fiducial points for the facial graph are extracted automatically by adjusting a grid of points to the result of an edge detector. After that, the position of the nodes, their relation with their neighbors and their Gabor jets are calculated in order to obtain the feature vector defining each face. A self-organizing map (SOM) framework is shown afterwards. Thus, the calculation of the winning neuron and the recognition process are performed by using a similarity function that takes into account both the geometric and texture information of the facial graph. The set of experiments carried out for our SOM-EBGM method shows the accuracy of our proposal when compared with other state-of the-art methods.

7

Recognizing the surgical situs in minimally invasive hip arthroplasty: A comparison of different filtering techniques

Putzer D., Pizzini A., Liebensteiner M., Moctezuma J. L., Nogler M.

Biocybernetics and Biomedical Engineering

|

2016

|

Vol. 36, no. 1

182--192

EN

Purpose: Detecting the soft tissue envelope and determining the work space available of the surgical situs during surgery is important for advanced instrument navigation techniques, wound care treatment, augmented reality, and instrument design. Different filtering techniques were evaluated to increase detectability of the soft tissue envelope. Methods: An algorithm was built for a time of flight (TOF) camera which recognizes the boarders of the soft tissue envelope. Different filtering techniques were tested on a dataset of eight surgical siti. Results: By using a median filter, a temporal filter and combining different input information provided by the time of flight camera by a logic operation the proposed algorithm was able to recognize the surgical situs in 73% of the images on average. Conclusions: The use of a TOF camera can introduce a new tool for recognizing the soft tissue envelope of a surgical approach.

8

A primal sub-gradient method for structured classification with the averaged sum loss

Mančev D., Todorović B.

International Journal of Applied Mathematics and Computer Science

|

2014

|

Vol. 24, no. 4

917--930

EN

We present a primal sub-gradient method for structured SVM optimization defined with the averaged sum of hinge losses inside each example. Compared with the mini-batch version of the Pegasos algorithm for the structured case, which deals with a single structure from each of multiple examples, our algorithm considers multiple structures from a single example in one update. This approach should increase the amount of information learned from the example. We show that the proposed version with the averaged sum loss has at least the same guarantees in terms of the prediction loss as the stochastic version. Experiments are conducted on two sequence labeling problems, shallow parsing and part-of-speech tagging, and also include a comparison with other popular sequential structured learning algorithms.

9

Detekcja znaków drogowych i kolejowych na zdjęciach i w chmurze punktów – przegląd istniejących algorytmów

Moskal A., Pastucha E.

Roczniki Geomatyki

|

2013

|

T. 11, z. 2(59)

69--78

PL

Badania i analizy możliwości automatycznego wykrywania znaków rozwijają się równolegle w wielu ośrodkach naukowych na świecie. Motywacje do prac zawierają się w większości w dwóch kategoriach: inwentaryzacja infrastruktury drogowej lub kolejowej oraz tworzenie systemów dla automatycznego wspomagania kierowcy. W zależności od wybranego kierunku, wykorzystywane są różnorakie dane pochodzące z różnych sensorów. Nowotworzone systemy wspomagania kierowcy wymagają sensorów o niewielkich gabarytach, dostarczających dane o małym rozmiarze, podczas gdy technologie tworzone na potrzeby inwentaryzacji znaków mogą korzystać z rozbudowanych systemów pomiarowych, integrujących różnorakie sensory pozyskujące bardzo dokładne, wysokorozdzielcze dane. Czas przetwarzania takich danych również zależy od potrzeb. Wykrycie i sklasyfikowanie znaku w systemach automatycznego wspomagania kierowcy musi być bardzo szybkie. Takich limitów nie trzeba stawiać przed systemami dla celów inwentaryzacji. Pozycjonowanie wykrywanych obiektów ma znaczenie jedynie w systemach inwentaryzujących, jednak nie jest wykluczone w pozostałych. Koncepcje algorytmów różnią się między ośrodkami badawczymi i wykorzystują wiele różnych nurtów w informatyce i matematyce. W artykule przedstawiono przegląd najważniejszych algorytmów z ostatnich piętnastu lat. Krótko opisano etapy pozyskania danych i systemy do tego wykorzystane. Następnie szeroko przedstawiono problem przygotowania danych, koncepcje wstępnego wykrycia znaków i ostatecznych klasyfikacji.

EN

During the last fifteen years, automatic sign recognition in different type of data has become the subject of many studies. Reasons for these works fall into one of two categories: inventory purposes or drivers assistance systems. Depending on the purpose of the systems, various types of sensors, acquiring different type of data, are implemented. Due to their application, drivers assistance systems need small sensors, bringing limited amount of data, while systems for inventory purposes can use complex measuring systems, integrating different types of sensors and providing high accuracy and large volume data. The time is also at issue. Detection and classification of a sign in driver assistance systems has to be done in real time, while processing of data for inventory purposes can be done off–line. Also global positioning of identified signs is significant only in the latter systems. Structures of proposed algorithms vary and use many different concepts, both from math and information processing. In this paper, basic concepts of most important algorithms from the last fifteen years are presented. Data acquisition process and measuring systems are described shortly. Then, data pre-processing, concepts of detection and, finally, concepts of classification are broadly covered.

10

Nonparametric methods of supervised classification

Jóźwik A.

Journal of Medical Informatics & Technologies

|

2013

|

Vol. 22

21--32

EN

Selected nonparametric methods of statistical pattern recognition are described. A part of them form modifications of the well known k-NN rule. To this group of the presented methods belong: a fuzzy k-NN rule, a pair-wise k-NN rule and a corrected k-NN rule. They can improve classification quality as compared with the standard k-NN rule. For the cases when these modifications would offer to large error rates an approach based on class areas determination is proposed. The idea of class areas can be also used for construction of the multistage classifier. A separate feature selection can be performed in each stage. The modifications of the k-NN rule and the methods based on determination class areas can be too slow in some applications, therefore algorithms for reference set reduction and condensation, for simple NN rule, are proposed. To construct fast classifiers it is worth to consider also a pair-wise linear classifiers. The presented idea can be used as in the case when the class pairs are linearly separable as well as in the contrary case.