Analyzing the constructive or destructive effects of the dependence of wireless communication channel coefficients on the coverage region (one of fundamental wireless performances) is of practical importance. Using Copula functions is one of the methods for modeling the dependence between interdependent random variables (RVs). In this paper, a wireless multiple-access fading channel (MAC) with independent sources and a coherent receiver is evaluated, where, the channel fading coefficients are assumed to be dependent on each other. Then, an exact expression for the coverage region is achieved using the Farlie-Gumbel-Morgenstern (FGM) Copula, and, it is proved that the negative dependence of the channel coefficients on each other has a constructive effect on the coverage region; that is, it expands the coverage region, but the positive dependency has a destructive effect on the coverage region. We present numerical illustrations to demonstrate the efficiency of the analytical results, validating our theoretical findings.
Micro-blogs and cyber-space social networks are the main communication mediums to receive and share news nowadays. As a side effect, however, the networks can disseminate fake news that harms individuals and the society. Several methods have been developed to detect fake news, but the majority require large sets of manually labeled data to attain the application-level accuracy. Due to the strict privacy policies, the required data are often inaccessible or limited to some specific topics. On the other side, quite diverse and abundant unlabeled data on social media suggests that with a few labeled data, the problem of detecting fake news could be tackled via semi-supervised learning. Here, we propose a semi-supervised self-learning method in which a sentiment analysis is acquired by some state-of-the-art pretrained models. Our learning model is trained in a semi-supervised fashion and incorporates LSTM with self-attention layers. We benchmark our model on a dataset with 20,000 news content along with their feedback, which shows better performance in precision, recall, and measures compared to competitive methods in fake news detection.
Graph learning and revealing the relationship between agents is an essential and pivotal problem in network science and machine learning, carrying significant implications across a broad range of applications. In the literature, there are generally two approaches to graph learning: model-based and learning-based methods. Recent learning-based methods like L2G, GLAD, and uGLAD add more flexibility to model structural priors in graph learning. We propose a novel approach, RL2G, which offers robustness to the "Learning to Learn Graph Topologies" method. This allows us to recover the topologies of graphs using a pre-trained model, independently from their size and topological properties. The experiments conducted on both synthetic and real-world data provide strong evidence that our approach performs well in revealing the relationships between different data entities.
The primary objective of medicine and technology is to provide services capable of identifying and treating patients based on their particular conditions. The accuracy of disease diagnosis is of paramount importance to this endeavor. Cancer is a major cause of death on a global scale, with prognosis, prevention, and prompt intervention offering the possibility of complete patient remission. In this Python-based experiment, we investigate the viability of transfer learning for lymph node diagnosis in histopathology images, which are used to fine-tune a DenseNet121-based pre-trained model. The results indicate that fine-tuning the DenseNet121 model is more effective than training the model from start. Due to variations in the network’s initial weight distribution, the network’s average accuracy is 96%. Following the implementation and completion of ten training epochs, the average overall accuracy reaches a maximum of 98%, with less than 10% error accuracy.
Automated facial expression recognition is popular for diagnosing mental diseases and interpreting human emotions. Image sentiment analysis helps incorporate feedback for improved product development. However, there's a lack of research on the effect of different color spaces on facial expression recognition performance. Thus, there are three objectives in this project, which are to develop a deep learningbased facial expression recognition system for image sentiment analysis. Second, to investigate the effect of different color spaces on facial expression recognition and select the best one. Lastly, to evaluate the accuracy of the system on the AffectNet dataset. The color spaces to be tested are RGB, YCBCR, HSV, XYZ, and YIQ. The methodology involves selecting the AffectNet dataset, pre-processing the images, building a Convolutional Neural Network model, and evaluating performance using metrics like accuracy, precision, recall, and F1-score.
