DSpace Arşivi :: by Yazar "Kaya, Yilmaz" değerine göre listeleniyor

Yazar "Kaya, Yilmaz" seçeneğine göre listele

Listeleniyor 1 - 20 / 76

1D-local binary pattern based feature extraction for classification of epileptic EEG signals
(Elsevier Science Inc, 2014) Kaya, Yilmaz; Uyar, Murat; Tekin, Ramazan; Yildirim, Selcuk
In this paper, an effective approach for the feature extraction of raw Electroencephalogram (EEG) signals by means of one-dimensional local binary pattern (1D-LBP) was presented. For the importance of making the right decision, the proposed method was performed to be able to get better features of the EEG signals. The proposed method was consisted of two stages: feature extraction by 1D-LBP and classification by classifier algorithms with features extracted. On the classification stage, the several machine learning methods were employed to uniform and non-uniform 1D-LBP features. The proposed method was also compared with other existing techniques in the literature to find out benchmark for an epileptic data set. The implementation results showed that the proposed technique could acquire high accuracy in classification of epileptic EEG signals. Also, the present paper is an attempt to develop a general-purpose feature extraction scheme, which can be utilized to extract features from different categories of EEG signals. (C) 2014 Elsevier Inc. All rights reserved.
A Computer Vision System for Classification of Some Euphorbia (Euphorbiaceae) Seeds Based on Local Binary Patterns
(IEEE, 2013) Kaya, Yilmaz; Karabacak, Osman; Caliskan, Abidin
In this study, a computer vision system was proposed for the seed images classification. The classification process was performed using uniform local binary patterns obtained from digital seed images. In this study, 240 (120 training and 120 test) images of the seed were used. First, the average uniform histograms of each type of seed (seed type classes) was obtained for the training set. Then the uniform LBP histogram of each seed in the test set were produced and compared with histograms of classes by using nearest neighbor. The Euclidean distance, sum square error, histogram intersection and Chi-square statistics were used to calculate the distance between seed samples. 95.83%. of seed images has been diagnosed properly with the proposed. As a result, the surface shape of the seeds include important information patterns to determine the taxonomic relationships, it is is expected that the computer vision systems provide significant advantages to identify the type of seed.
A Computer Vision System for the Automatic Identification of Butterfly Species via GaborFilter-Based Texture Features and Extreme Learning Machine: GF+ELM
(Assoc Information Communication Technology Education & Science, 2013) Kaya, Yilmaz; Kayci, Lokman; Tekin, Ramazan
Butterflies are classified first according to their outer morphological qualities. It is required to analyze their genital characters when classification according to their outer morphological qualities is not possible. The genital characters of butterflies can be obtained using various chemical substances and methods; however, these processes can only be carried out with some certain expenses. Furthermore, the preparation of genital slides is time-consuming since it requires specific processes. In this study, a new method based on the extreme learning machine (ELM) and Gabor filters (GFs), which is an image processing technique, was used for the identification of butterfly species as an alternative to conventional diagnostic methods. GFs have been recognized as a very useful tool in texture analysis, due to their optimal localization properties in both the spatial and frequency domains, and have been found to have widespread use in computer vision applications. To obtain the appropriate features from butterfly images in the spatial domain, 20 filters were designed for the various angles and frequencies (5 frequencies and 4 orientations). The diagnosing of butterflies was performed through ELM, with texture features based on GFs. The classification process was performed with a 75%-25% training-test set for different activation functions and the recognition performance value was obtained as 97.00%. In addition, the recognition success rates with ELM were compared to other machine learning methods and it was seen that ELM has a more significant success rate in butterfly identification than other methods. As a result, the proposed method is a suitable machine vision system for detecting butterfly species
A New Approach for Congestive Heart Failure and Arrhythmia Classification Using Angle Transformation with LSTM
(Springer Heidelberg, 2022) Kaya, Yilmaz; Kuncan, Fatma; Tekin, Ramazan
Electrocardiogram (ECG) is widely used as a diagnostic method to identify various heart diseases such as heart failure, cardiac and sinus rhythms. The ECG signal analyzes the electrical activity of the heart and shows waveforms that help detect heart irregularities. A new approach is suggested for automatic identification of congestive heart failure (CHF) and arrhythmia (ARR). In this study, long short-term memory neural networks (LSTM) were used to classify ECG signals by combining LSTM and angle transform (AT) methods. The AT uses the angular information of the neighbor signals on both sides of the target signal to classify ECG signals. The new signals obtained as a result of AT conversion vary between 0 and 359. Histogram of new signals determines the inputs to the LSTM method. LSTM uses histograms to distinguish between three different conditions: ARR, CHF, and normal sinus rhythm (NSR). The proposed approach is tested on ECG signals received from MIT-BIH and BIDMC databases. The experimental results have shown that the proposed method, AT + LSTM, has achieved high success rate of classifying ECG signals. The success rate in classifying CHF, ARR, and NSR ECG signals for 70-30% training sets was observed as 98.97%. Further experiments were conducted for varying training-testing dataset ratio to demonstrate the robustness of the proposed approach, and success rates are observed between 98.56 and 100%. Another experiment regarding different values of the dR and dL distance parameters of the AT model has shown that the performance of the proposed method increases while increasing the distance value. The success rates from increasing the distance value were obtained between 98.97 and 100%. To show the effect of segment lengths of ARR, NSR, and CHF signals on classification success, these signals were divided into segments of 10,000, 5000, and 1000 lengths. Achieved success rates ranged from 97.75 to 98.97%. Considering the results, high results were observed with the AT + LSTM approach, which is generally recommended in all scenarios.
A new approach for congestive heart failure and arrhythmia classifiication using downsampling local binary patterns with LSTM
(Tubitak Scientific & Technological Research Council Turkey, 2022) Akda, Sueleyman; Kuncan, Fatma; Kaya, Yilmaz
Electrocardiogram (ECG) is a vital diagnosis approach for the rapid explication and detection of various heart diseases, especially cardiac arrest, sinus rhythms, and heart failure. For this purpose, in this study, a different perspective based on downsampling one-dimensional-local binary pattern (1D-DS-LBP) and long short-term memory (LSTM) is presented for the categorization of Electrocardiogram (ECG) signals. A transformation method named 1D-DS-LBP has been presented for Electrocardiogram signals. The 1D-DS-LBP method processes the bigness smallness relationship between neighbors. According to the proposed method, by downsampling the signal, the histograms of 1D local binary patterns (1D-LBP) calculated from the obtained signal groups are collected and included as a reference to the long short-term memory structure. The long short-term memory structure has been applied to 1D-DS-LBP conversion applied ECG signals with both unidirectional and bidirectional. To test the proposed approach, ECG signals of three (3) different states of congestive heart failure (CHF), arrhythmia (ARR), and normal sinus rhythm (NSR) consisting of 972 signals were used. Signals were taken from the MIT-BIH and BIDMC databases. Experiments were carried out in various scenarios. We observed that the success rate of the proposed approach obtained very high classification accuracies compared to other studies in the literature. The obtained ECG diagnostic performance values varied between 96.80% and 99.79%. Based on this, this approach has a high potential to have a wide field of study in medical applications.
A NEW APPROACH FOR DIAGNOSTIC ESTIMATION OF OBSTRUCTIVE SLEEP APNEA SYNDROME BASED ON ONE DIMENSIONAL LOCAL BINARY PATTERN
(IEEE, 2014) Kaya, Yilmaz; Sezgin, Necmettin; Tekin, Ramazan
In this study, a new approach for estimation of Obstructive Sleep Apnea Syndrome (OSAS) was proposed. OSAS is a sleep disorder that affects the life comfortability in human life. Up to now, the OSAS was diagnosed by Polysomnography (PSG) device by connected to the patients via electrodes. This device is expensive and restricted in the clinics. Since OSAS is serious, it should be diagnosed and treated early. For this purpose, the recorded Electroencephalography (EEG), Electromyography (EMG) and snore data were analyzed and features of them extracted by a proposed method called One Dimensional Local Binary Pattern (1D-LBP). The 1D-LBP extracted features from raw data effectively. The features, then, were fed to classifier's input in order to diagnose OSAS. As a result most of tested classifiers have yielded accuracies over 99%. The best results were obtained by using EEG, EMG and snore signal altogether. It was also shown that while the complexity of signal increase the best accuracy was obtained at the output of the classifier. The results have shown that the 1D-LBP method is an acceptable and has advantageous over conventional methods due to its capable of extract significant features from more complex signal. The results can be used in sleep laboratory for help to experts before put patient to the PSG.
A new approach for diagnostic estimation of Obstructive Sleep Apnea Syndrome based on One Dimensional Local Binary Pattern
(IEEE Computer Society, 2014) Kaya, Yilmaz; Sezgin, Necmettin; Tekin, Ramazan
In this study, a new approach for estimation of Obstructive Sleep Apnea Syndrome (OSAS) was proposed. OSAS is a sleep disorder that affects the life comfortability in human life. Up to now, the OSAS was diagnosed by Polysomnography (PSG) device by connected to the patients via electrodes. This device is expensive and restricted in the clinics. Since OSAS is serious, it should be diagnosed and treated early. For this purpose, the recorded Electroencephalography (EEG), Electromyography (EMG) and snore data were analyzed and features of them extracted by a proposed method called One Dimensional Local Binary Pattern (1D-LBP). The 1D-LBP extracted features from raw data effectively. The features, then, were fed to classifier's input in order to diagnose OSAS. As a result most of tested classifiers have yielded accuracies over 99%. The best results were obtained by using EEG, EMG and snore signal altogether. It was also shown that while the complexity of signal increase the best accuracy was obtained at the output of the classifier. The results have shown that the 1D-LBP method is an acceptable and has advantageous over conventional methods due to its capable of extract significant features from more complex signal. The results can be used in sleep laboratory for help to experts before put patient to the PSG. © 2014 IEEE.
A New Approach for Human Recognition Through Wearable Sensor Signals
(Springer Heidelberg, 2021) Kilic, Safak; Kaya, Yilmaz; Askerbeyli, Iman
Recently, subjects such as human recognition (HR), age estimation and gender recognition have been among the most investigated human-computer interaction topics, in both the academic and other fields. HR is a process in which a person is detected based on the obtained biometrical features. In this study, a new feature extraction method has been suggested through using the signals received from the sensors of the accelerometer, magnetometer and gyroscope attached to the 5 areas on the human body. The feature extraction from the signals is one of the most crucial stage. The reason behind the success of HR is based on the extracted features. However, the extraction of appropriate features for HR is a challenging issue. Various transformation methods like 1D-LBP and 1D-FbLBP have been applied to the sensor-based signals. Following the transformation process, the statistical features have been acquired from the newly developed signals. The classification processes have been carried out with the distinctive methods concerning machine learning (Knn, RF, A1DE, A2DE and ANN) by using these features. According to these results, 1D-LBP (88.4649%) and 1D-FbLBP (91.8281%) methods have been chosen to provide effective features for HR.
A new approach for physical human activity recognition based on co-occurrence matrices
(Springer, 2022) Kuncan, Fatma; Kaya, Yilmaz; Tekin, Ramazan; Kuncan, Melih
In recent years, it has been observed that many researchers have been working on different areas of detection, recognition and monitoring of human activities. The automatic determination of human physical activities is often referred to as human activity recognition (HAR). One of the most important technology that detects and tracks the activity of the human body is sensor-based HAR technology. In recent days, sensor-based HAR attracts attention in the field of computers due to its wide use in daily life and is a rapidly growing field of research. Activity recognition (AR) application is carried out by evaluating the signals obtained from various sensors placed in the human body. In this study, a new approach is proposed to extract features from sensor signals using HAR. The proposed approach is inspired by the Gray Level Co-Occurrence Matrix (GLCM) method, which is widely used in image processing, but it is applied to one-dimensional signals, unlike GLCM. Two datasets were used to test the proposed approach. The datasets were created from the signals obtained from the accelerometer, gyro and magnetometer sensors. Heralick features were obtained from co-occurrence matrix created after 1D-GLCM (One (1) Dimensional-Gray Level Co-Occurrence Matrix) was applied to the signals. HAR operation has been carried out for different scenarios using these features. Success rates of 96.66 and 93.88% were obtained for two datasets, respectively. It has been observed that the new approach proposed within the scope of the study provides high success rates for HAR applications. It is thought that the proposed approach can be used in the classification of different signals.
A new approach for physical human activity recognition from sensor signals based on motif patterns and long-short term memory
(Elsevier Sci Ltd, 2022) Kuncan, Fatma; Kaya, Yilmaz; Yiner, Zueleyha; Kaya, Mahmut
Numerous studies have been carried out in recent years on the recognition, tracking, and discrimination of human activities. Automatic recognition of physical activities is often referred to as human activity recognition (HAR). There are generally vision-based and sensor-based approaches for activity recognition. The computer vision-based approach generally works well in laboratory conditions, but it can fail in real-world problems due to clutter, variable light intensity, and contrast. Sensor-based HAR systems are realized by continuously monitoring and analyzing physiological signals measured from heterogeneous sensors connected to the person's body. In this study, the Motif Patterns (MP) approach, which extracts features from sensor signals, is proposed for HAR. The success of the HAR systems depends on the effectiveness of the features extracted from the signals. The LSTM network is a special kind of recurrent neural network that has been used to make very successful predictions on time series data where long-term dependencies are. The LSTM network type offers a successful solution approach to solving long-term dependencies problems such as human activity recognition. The classification process was carried out with Long-Short Term Memory (LSTM) using MP features extracted from accelerometer, gyroscope, and magnetometer sensor signals. A large dataset of 9120 signals was used to test the proposed approach. A high success rate of 98.42 % was achieved with the proposed MP + LSTM method. As a result, it has been seen that the proposed approach has been obtained with a high success rate for HAR using sensor signals.
A new approach for remaining useful life prediction of bearings using 1D-ternary patterns with LSTM
(Springer Heidelberg, 2023) Akcan, Eyyup; Kaya, Yilmaz
Bearings frequently experience malfunctions in mechanical systems, directly impacting system performance. Accurate prediction of bearing failures is crucial for maintenance planning and preventing unexpected system breakdowns. Data-driven prognostic techniques are commonly employed to estimate the remaining useful life (RUL) of high-speed bearings. RUL prediction relies on establishing the fundamental relationship between bearing degradation and its current health status, with the accuracy depending on effective feature extraction from the bearing data. In this study, a novel approach is proposed for the RUL prediction of bearings. The 1D-TP method is applied to vibration signals, resulting in two feature vectors, LOWER and UPPER, which are then utilized in combination with LSTM for RUL prediction. The proposed approach is evaluated using a dataset from the PRONOSTIA platform, and performance metrics including MAE, RMSE, SMAPE, RA, and Score are determined. The results demonstrate that the 1D-TP + LSTM method successfully predicts the remaining life of bearings. Accurate RUL assessment and reliability analysis aid personnel in making informed maintenance decisions, preventing losses from mechanical system damage, improving production safety, and safeguarding the mechanical system from harm.
A new approach to COVID-19 detection from x-ray images using angle transformation with GoogleNet and LSTM
(Iop Publishing Ltd, 2022) Kaya, Yilmaz; Yiner, Zuleyha; Kaya, Mahmut; Kuncan, Fatma
Declared a pandemic disease, COVID-19 has affected the lives of millions of people and had significant effects on public health. Despite the development of effective vaccines against COVID-19, cases continue to increase worldwide. According to studies in the literature, artificial intelligence methods are used effectively for the detection of COVID-19. In particular, deep-learning-based approaches have achieved very good results in clinical diagnostic studies and other fields. In this study, a new approach using x-ray images is proposed to detect COVID-19. In the proposed method, the angle transform (AT) method is first applied to the x-ray images. The AT method proposed in this study is an important novelty in the literature, as there is no such approach in previous studies. This transformation uses the angle information created by each pixel on the image with the surrounding pixels. Using the AT approach, eight different images are obtained for each image in the dataset. These images are trained with a hybrid deep learning model, which combines GoogleNet and long short-term memory (LSTM) models, and COVID-19 disease detection is carried out. A dataset from the Mendeley database is used to test the proposed approach. A high classification accuracy of 98.97% is achieved with the AT + GoogleNet + LSTM approach. The results obtained were also compared with other studies in the literature. The presented results reveal that the proposed method is successful for COVID-19 detection using chest x-ray images. Direct transfer methods were also applied to the data set used in the study. However, worse results were observed according to the proposed approach. The proposed approach has the flexibility to be applied effectively to different medical images.
A new automatic bearing fault size diagnosis using time-frequency images of CWT and deep transfer learning methods
(Tubitak Scientific & Technological Research Council Turkey, 2022) Kaya, Yilmaz; Kuncan, Fatma; Ertunc, H. Metin
Bearings are generally used as bearings or turning elements. Bearings are subjected to high loads and rapid speeds. Furthermore, metal-to-metal contact within the bearing makes it sensitive. In today's machines, bearing failures disrupt the operation of the system or completely stop the system. Bearing failures that can occur can cause enormous damage to the entire system. Therefore, it is necessary to anticipate bearing failures and to carry out a regular diagnostic examination. Various systems have been developed for fault diagnosis. In recent years, deep transfer learning (DTL) methods are often preferred in current bearing diagnosis models, as they provide time savings and high success rates. Deep transfer learning models also improve diagnosis accuracy under certain conditions by greatly reducing human intervention. Diagnosis at the right time is very important for the sustainability and efficiency of industrial production. A technique based on continuous wavelet transform (CWT) and two dimensional (2D) convolutional neural networks (CNN) is presented in this paper to detect fault size from vibration data of various bearing failure types. Time-frequency (TF) color scalogram images for bearing vibration signals were obtained using the CWT method. Using AlexNet, GoogleNet, Resnet, VGG16, and VGG19 deep transfer learning methods with scalogram images, fault size prediction from vibration signals was performed. Five different transfer deep learning models were used for three different data sets. It was observed that the success rates obtained varied between 96.67% and 100%.
A new content-free approach to identification of document language: Angle patterns
(Gazi Univ, Fac Engineering Architecture, 2022) Noyan, Tuba; Kuncan, Fatma; Tekin, Ramazan; Kaya, Yilmaz
Graphical/Tabular Abstract Language identification (LI) in text mining is the process of detecting the natural language in which a document or part of it is written. LI aims to mimic a human's ability to recognize certain languages from text by computer algorithms. LI can be defined as a classification problem subject based on the information used in word or character size for any document. When the literature is examined for LI application, it is seen that various linguistic or statistical-based approaches are used. Linguistic methods are methods that perform LI according to a special word or character of a language. These methods are applied based on the special rules of the languages. When we look at the statistical methods, it shows that the words or characters that make up the language depend on their frequency and distribution. The statistical approaches used are content -independent methods. The semantic context of the text is not concerned with its content. According to linguistic methods, it does not provide sufficient information about the content of the text. The proposed model in this study is a statistical approach. Figure A. Proposed block diagram for LI Purpose: In this study, a new LI approach using the angle information between the UTF-8 values of the characters in the text is proposed. The proposed angle pattern method is used for feature extraction from texts. Angle patterns method is a statistical approach. In the angle method, there are two distance parameters, R and L, which express which neighborhood to look at from the reference point to the left and right. Theory and Methods: To test the proposed approach, four datasets, two created by the authors and two publicly available on the Internet, were used. By using the features obtained by the angle pattern method, classification process was carried out with different machine learning methods such as Random Forest, Support Vector Machine, Linear Discriminant Analysis, Naive Bayes and K-nearest neighbor. Language identification performance results determined from four different data sets were observed as 96.81%, 99.39%, 93.31% and 98.60%, respectively. Results: According to the performance results achieved as a result of the study, it has been determined that the proposed angle pattern method provides important distinguishing information in language identification application. It is thought that the proposed approach in this study can be used in many different text mining applications such as spam recognition, text categorization, as well as LI application.
A new feature extraction approach based on one dimensional gray level co-occurrence matrices for bearing fault classification
(Taylor & Francis Ltd, 2021) Kaya, Yilmaz; Kuncan, Melih; Kaplan, Kaplan; Minaz, Mehmet Recep; Ertunc, H. Metin
Recently, precise and deterministic feature extraction is one of the current research topics for bearing fault diagnosis. For this aim, an experimental bearing test setup was created in this study. In this setup, vibration signals were obtained from the bearings on which artificial faults were generated in specific sizes. A new feature extraction method based on co-occurrence matrices for bearing vibration signals was proposed instead of the conventional feature extraction methods, as in the literature. The One (1) Dimensional-Local Binary Patterns (1D-LBP) method was first applied to bearing vibration signals, and a new signal whose values ranged between 0-255 was obtained. Then, co-occurrence matrices were obtained from these signals. The correlation, energy, homogeneity, and contrast features were extracted from these matrices. Different machine learning methods were employed with these features to carry out the classification process. Three different data sets were used to test the proposed approach. As a result of analysing the signals with the proposed model, the success rate is 87.50% for dataset1 (different speed), 96.5% for dataset2 (fault size (mm)) and 99.30% for dataset3 (fault type - inner ring, outer ring, ball) was found, respectively.
A new intelligent classifier for breast cancer diagnosis based on a rough set and extreme learning machine: RS plus ELM
(Tubitak Scientific & Technological Research Council Turkey, 2013) Kaya, Yilmaz
Breast cancer is one of the leading causes of death among women all around the world. Therefore, true and early diagnosis of breast cancer is an important problem. The rough set (RS) and extreme learning machine (ELM) methods were used collectively in this study for the diagnosis of breast cancer. The unnecessary attributes were discarded from the dataset by means of the RS approach. The classification process by means of ELM was performed using the remaining attributes. The Wisconsin Breast Cancer dataset (WBCD), derived from the University of California Irvine machine learning database, was used for the purpose of testing the proposed hybrid model and the success rate of the RS + ELM model was determined as 100%. Moreover, the most appropriate attributes for the diagnosis of breast cancer were determined from the WBCD in this study. It is considered that the proposed method will be useful in similar medical practices.
A new local pooling approach for convolutional neural network: local binary pattern
(Springer, 2023) Ozdemir, Cuneyt; Dogan, Yahya; Kaya, Yilmaz
The pooling layer used in CNN models aims to reduce the resolution of image/feature maps while retaining their distinctive information, reducing computation time and enabling deeper models. Max and average pooling methods are frequently used in CNN models due to their computational efficiency; however, these methods discard the position information of the pixels. In this study, we proposed an LBP-based pooling method that generates a neighborhood-based output for any pixel, reflecting the correlation between pixels in the local area. Our proposed method reduces information loss since it considers the neighborhood and size of the pixels in the pooling region. Experimental studies were performed on four public datasets to assess the effectiveness of the LBP pooling method. In experimental studies, a toy CNN model and various transfer learning models were utilized in conducting test operations. The proposed method provided improvements of 1.56% for Fashion MNIST, 0.22% for MNIST, 3.95% for CIFAR10, and 5% for CIFAR100 dataset using the toy model. In the experimental studies conducted using the transfer learning model, performance improvements of 6.99(-/+)(0.74) and 8.3(-/+)(0.1) were achieved for CIFAR10 and CIFAR100, respectively. We observed that the proposed method outperforms the commonly used pooling layers in CNN models. Code for this paper can be publicly accessed at: https://github.com/cuneytozdemir/lbppooling
A Novel Approach for Activity Recognition with Down-Sampling 1D Local Binary Pattern Features
(Univ Suceava, Fac Electrical Eng, 2019) Kuncan, Fatma; Kaya, Yilmaz; Kuncan, Melih
The sensors on the mobile devices directly reflect the physical and demographic characteristics of the user. Sensor signals may contain information about the gender and movement of the person. Automatic recognition of physical activities often referred to as human activity recognition (HAR). In this study, a novel feature extraction approach for the HAR system using the mobile sensor signals, the Down Sampling One Dimensional Local Binary Pattern (DS-1D-LBP) method is proposed. Feature extraction from signals is one of the most critical stages of HAR because the success of the HAR system depends on the features extraction. The proposed HAR system consists of two stages. In the first stage, DS-ID-LBP conversion was applied to the sensor signals in order to extract statistical features from the newly formed signals. In the last stage, classification with Extreme Learning Machine (ELM) was performed using these features. The highest success rate was 96.87 percent in the experimental results according to the different parameters of DS-ID-LBP and ELM. As a result of this study, the novel approach demonstrated that the proposed model performed with a high success rate using mobile sensor signals for the HAR system.
A novel approach for SEMG signal classification with adaptive local binary patterns
(Springer Heidelberg, 2016) Ertugrul, Omer Faruk; Kaya, Yilmaz; Tekin, Ramazan
Feature extraction plays a major role in the pattern recognition process, and this paper presents a novel feature extraction approach, adaptive local binary pattern (aLBP). aLBP is built on the local binary pattern (LBP), which is an image processing method, and one-dimensional local binary pattern (1D-LBP). In LBP, each pixel is compared with its neighbors. Similarly, in 1D-LBP, each data in the raw is judged against its neighbors. 1D-LBP extracts feature based on local changes in the signal. Therefore, it has high a potential to be employed in medical purposes. Since, each action or abnormality, which is recorded in SEMG signals, has its own pattern, and via the 1D-LBP these (hidden) patterns may be detected. But, the positions of the neighbors in 1D-LBP are constant depending on the position of the data in the raw. Also, both LBP and 1D-LBP are very sensitive to noise. Therefore, its capacity in detecting hidden patterns is limited. To overcome these drawbacks, aLBP was proposed. In aLBP, the positions of the neighbors and their values can be assigned adaptively via the down-sampling and the smoothing coefficients. Therefore, the potential to detect (hidden) patterns, which may express an illness or an action, is really increased. To validate the proposed feature extraction approach, two different datasets were employed. Achieved accuracies by the proposed approach were higher than obtained results by employed popular feature extraction approaches and the reported results in the literature. Obtained accuracy results were brought out that the proposed method can be employed to investigate SEMG signals. In summary, this work attempts to develop an adaptive feature extraction scheme that can be utilized for extracting features from local changes in different categories of time-varying signals.
A novel approach for spam email detection based on shifted binary patterns
(Wiley-Hindawi, 2016) Kaya, Yilmaz; Ertugrul, Omer Faruk
Advances in communication allow people flexibility to communicate in various ways. Electronic mail (email) is one of the most used communication methods for personal or business purposes. However, it brings one of the most tackling issues, called spam email, which also raises concerns about data safety. Thus, the requirement of detecting spams is crucial for keeping the users safe and saving them from the waste of time while tackling those issues. In this study, an effective approach based on the probability of the usage of the characters that has similar orders with respect to their UTF-8 value by employing shifted one-dimensional local binary pattern (shifted-1D-LBP) was used to extract quantitative features from emails for spam email detection. Shifted-1D-LBP, which can be described as an ordered set of binary comparisons of the center value with its neighboring values, is a content-based approach to spam detection with low-level information. To validate the performance of the proposed approach, three benchmark corpora, Spamassasian, Ling-Spam, and TREC email corpuses, were used. The average classification accuracies of the proposed approach were 92.34%, 92.57%, and 95.15%, respectively. Analysis and promising experimental results indicated that the proposed approach was a very competitive feature extraction method in spam email filtering. Copyright (c) 2016 John Wiley & Sons, Ltd.

Yazar "Kaya, Yilmaz" seçeneğine göre listele

Sayfa Başına Sonuç

Sıralama seçenekleri