Yazar "Aslan, Muzaffer" seçeneğine göre listele

Listeleniyor 1 - 4 / 4
  • Küçük Resim
    Öğe
    Classification of physical actions from surface EMG signals using the wavelet packet transform and local binary patterns
    (Institute of Physics Publishing, 2020) Alçin, Ömer Faruk; Budak, Ümit; Aslan, Muzaffer; Akbulut, Yaman; Cömert, Zafer; Akpınar, Muhammed H.; Şengür, Abdulkadir
    Physical action recognition is a hot topic in human-machine interactions. It has potential uses in helping disabled people and in various robotic applications. Electromyography (EMG) signals measure the electrical activity of the muscular systems involved in physical actions. In this chapter, an efficient approach is developed for physical action recognition in humans based on EMG signals. The proposed method is composed of signal decomposition, feature extraction and feature classification. The signal decomposition is carried out using the wavelet packet transform (WPT). The WPT successively decomposes an input signal into its approximation and detail coefficients, which offers a more productive signal analysis. A one-dimensional local binary pattern (LBP) is used to code the approximation and detail coefficients of the decomposed EMG signals. The histogram of the LBP is used as the feature vectors of the EMG physical action classes. The support vector machine (SVM), decision tree, linear discriminant, k-nearest neighbors (k-NN), boosted and bagged tree ensemble classifiers are used in the classification stage. A dataset taken from the UCI machine learning repository is used in the experiments. The Delsys EMG apparatus, which has eight electrodes, is used to record the surface EMG signals. Each class of the dataset contains three male subjects and one female subject. The dataset contains ten physical actions, namely hugging, jumping, bowing, clapping, handshaking, running, sitting, standing, walking and waving. The experiments are carried out on each electrode with a ten-fold cross-validation test, and the average accuracy score is calculated. The experimental results show that the proposed method is quite efficient in EMG signal classification. The calculated average accuracy is 100% for each electrode.
  • Küçük Resim
    Öğe
    Deep rhythm and long short term memory-based drowsiness detection
    (Elsevier, 2021) Türkoğlu, Muammer; Alçin, Ömer Faruk; Aslan, Muzaffer; Al-Zebari, Adel; Şengur, Abdülkadir
    In this paper, a deep-rhythm-based approach is proposed for the efficient detection of drowsiness based on EEG recordings. In the proposed approach, EEG images are used instead of signals where the time and frequency information of the EEG signals are incorporated. The EEG signals are converted to EEG images using the time-frequency transformation method. The Short-Time-Fourier-Transform (STFT) is used for this transformation due to its simplicity. The rhythm images are then extracted by dividing the EEG images based on frequency intervals. EEG signals contain five rhythms, namely Delta rhythm (0–4?Hz), Theta rhythm (4–8?Hz), Alpha rhythm (8–12?Hz), Beta rhythm (12–30?Hz), and Gamma rhythm (30–50?Hz). From each rhythm image, deep features are extracted based on a pre-trained convolutional neural network (CNN) model, with pre-trained residual network (ResNet) models such as ResNet18, ResNet50, and ResNet101. The obtained deep features from each rhythm image are fed into the Long-Short-Term-Memory (LSTM) layer, and the LSTM layers are then sequentially connected to each other. After the last LSTM layer, a fully-connected layer, a softmax layer, and a classification layer are employed in order to detect the class labels of the input EEG signals. Various experiments were conducted with the MIT/BIH Polysomnographic Dataset. The experiments showed that the concatenated ResNet features achieved an accuracy score of 97.92%. The obtained accuracy score was also compared with the state-of-the-art scores and, to the best of our knowledge, the proposed method achieved the best accuracy score among the methods compared.
  • Küçük Resim
    Öğe
    Electrocardiogram beat classification using deep convolutional neural network techniques
    (Institute of Physics Publishing, 2020) Cömert, Zafer; Akbulut, Yaman; Akpınar, Muhammed H.; Alçin, Ömer Faruk; Budak, Ümit; Aslan, Muzaffer; Şengür, Abdulkadir
    The electrocardiogram (ECG) is a useful method which enables the monitoring of various cardiac conditions, such as arrhythmia and heart rate variability (HRV). ECG beats help to determine various heart failures such as cardiac disease and ventricular tachyarrhythmia. In the literature, it can be seen that various advanced signal processing and machine learning techniques and deep learning algorithms have been employed for ECG beat categorization. These methods were generally based on either the time domain or frequency domain. Time-frequency (T-F) based techniques have also been proposed for ECG beat classification. In this chapter, a different model is proposed for the ECG beat classification task. In the proposed approach, the ECG beats are initially represented by images. Instead of using a time-frequency approach for converting the ECG beats to ECG images, we opt to use the ECG beats directly to construct the ECG images. In other words, the ECG beat values are directly saved as ECG images. Three deep convolutional neural network (CNN) approaches are considered in ECG beat classification. These approaches ensure end-to-end learning schema, fine-tuning of pre-trained CNN models, extraction of deep features and their classification using a traditional classifier, such as the support vector machine (SVM) or deep machine learning approaches. The well-known MIT-BIH arrhythmia database is considered in the evaluation of the proposed deep learning approaches. The database is separated into two sets, the training and test dataset in proportions of 75% and 25%, respectively. The experimental results are evaluated using the classification accuracy score. The results show that the proposed methods have potential for use in ECG beat classification.
  • Küçük Resim Yok
    Öğe
    Wavelet ELM-AE Based Data Augmentation and Deep Learning for Efficient Emotion Recognition Using EEG Recordings
    (Ieee-Inst Electrical Electronics Engineers Inc, 2022) Ari, Berna; Siddique, Kamran; Alcin, Omer Faruk; Aslan, Muzaffer; Sengur, Abdulkadir; Mehmood, Raja Majid
    Emotion perception is critical for behavior prediction. There are many ways to capture emotional states by observing the body and copying actions. Physiological markers such as electroencephalography (EEG) have gained popularity, as facial emotions may not always adequately convey true emotion. This study has two main aims. The first is to measure four emotion categories using deep learning architectures and EEG data. The second purpose is to increase the number of samples in the dataset. To this end, a novel data augmentation approach namely the Extreme Learning Machine Wavelet Auto Encoder (ELM-W-AE) is proposed for data augmentation. The proposed data augmentation approach is both simple and faster than the other synthetic data augmentation approaches. For deep architectures, large datasets are important for performance. For this reason, data multiplexing approaches with classical and synthetic methods have become popular recently. The proposed synthetic data augmentation is the ELM-W-AE because of its efficiency and detail reproduction. The ELM-AE structure uses wavelet activation functions such as Gaussian, groove gap waveguide (GGW), Mexican, Meyer, Morlet, and Shannon. Deep convolutional architectures classify EEG signals as images. EEG waves are scalograms using Continuous Wavelet Transform (CWT). The ResNet18 architecture recognizes emotions. The proposed technique uses GAMEEMO data collected during gameplay. Each of these states is represented in the GAMEEMO data collection. The visual data set created from the signal was divided into two groups 70% training and 30% testing. ResNet18 has been fine-tuned with augmented photos, training images only. It achieved 99.6% classification accuracy in tests. The proposed method is compared with the other approaches on the same dataset, and an approximately 22% performance improvement is achieved.