|
|
|
|
|
 |
Search published articles |
 |
|
Showing 2 results for Classification
Sina Shamekhi, Mohammad Fouladvand, Ali Ahmad Alipour,
Volume 9, Issue 1 (9-2022)
Abstract
Nowadays, sleep deprivation is a pervasive problem that affects human physical and mental health. In this research, the effects of sleep deprivation on brain function and its diagnosis have been studied using electroencephalogram (EEG) signals recorded from 30 subjects after complete sleep and one day of sleep deprivation with open and closed eyes. Linear features like signal power and nonlinear features consisting of Shannon, Renyi, sample, and permutation entropies were extracted from signals. We used the PCA algorithm and Wilcoxon feature ranking method to extract the superior features and employed SVM, KNN, and a Decision tree to detect sleep-deprived cases. Brain maps of extracted features were plotted using the sLORETA algorithm to investigate the effects of sleep deprivation. Based on the results, the decision tree classifier with 100 superior selected features of Wilcoxon achieved the best performance with accuracy and precision of 99.0% and 99.8%, respectively. Also, comparing the results of linear and nonlinear features reveals the impressive role of the nonlinear features in the classification problem of this work. The maps of the features revealed noticeable changes in the level of attention, concentration, decision-making, and visual and movement activities.
Hamed Riazati Seresht, Dr. Karim Mohammadi,
Volume 10, Issue 1 (3-2023)
Abstract
Insufficient training data is one of the main challenges of utilizing deep Convolutional Neural Networks (CNNs) for Environmental Sound Classification (ESC). As a promising solution, Transfer Learning (TL) has addressed this issue by adapting a network pre-trained on a large-scale dataset to the target task. In this paper, we demonstrate that not all neurons/kernels of every layer in CNN networks are equally utilized to process the inputs of different classes, but there is a specific subgroups of neurons/kernels in every layer that play the key role in classification of every output class. Based on this observation and due to similarities that exist between feature spaces of some source and target classes, we propose to concentrate the fine-tuning process only on those neurons/kernels that do need changes and have the greatest impact on misclassifying target data. To identify these neurons/kernels, we pose a nested optimization problem for which we propose an effective evolutionary approach as solution. Compared to the conventional fine-tuning approach, our proposed method achieves absolute improvements of about 1.9% and 2.3% in accuracy on ESC-50 and DCASE-17, respectively; remarkable improvements produced not by adding augmented data but with a more efficient utilization of knowledge stored in the pre-trained network. It is noteworthy that the computation time overhead of the proposed evolutionary method is rather small (about one third of the time required to train the model from scratch.
|
|
|
|
|
| نشریه سامانههای غیرخطی در مهندسی برق در خصوص اصول اخلاقی انتشار مقاله، از توصیههای «کمیته بینالمللی اخلاق نشر» موسوم به COPE و «منشور و موازین اخلاق پژوهش» مصوب معاونت پژوهش و فناوری وزارت علوم، تحقیقات و فناوری تبعیت میکند. |
|
|
|
|
|
