Diabetic Retinopathy (DR) is a result of prolonged diabetes with poor blood sugar management. It causes vision problems and blindness due to the deformation of the human retina. Recently, DR has become a crucial medical problem that affects the health and life of people. Diagnosis of DR can be done manually by ophthalmologists, but this is cumbersome and time consuming especially in the current overloaded physician’s environment. The early detection and prevention of DR, a severe complication of diabetes that can lead to blindness, require an automatic, accurate, and personalized machine learning-based method. Various deep learning algorithms, particularly convolutional neural networks (CNNs), have been investigated for detecting different stages of DR. Recently, transformers have proved their capabilities in natural language processing. Vision transformers (ViTs) are extensions of these models to capture long-range dependencies in images, which achieved better results than CNN models. However, ViT always needs huge datasets to learn properly, and this condition reduced its applicability in DR domain. Recently, a new real-world and large fundus image dataset called fine-grained annotated diabetic retinopathy (FGADR) has been released which supported the application of ViT in DR diagnosis domain. The literature has not explored FGADR to optimize ViT models. In this paper, we propose a novel ViT based deep learning pipeline for detecting the severity stages of DR based on fundus photography-based retina images. The model has been built using FGADR dataset. The model has been optimized using a new optimizer called AdamW to detect the global context of images. Because FGADR is an imbalanced dataset, we combine several techniques for handling this issue including the usage of F1-score as the optimization metric, data augmentation, class weights, label smoothing, and focal loss. Extensive experiments have been conducted to explore the role of ViT with different data balancing techniques to detect DR. In addition, the proposed model has been compared with the state-of-the-art CNN algorithms such as ResNet50, Incep-tionV3, and VGG19. The adopted model was able to capture the crucial features of retinal images to understand DR severity better. It achieved superior results compared to other CNN and baseline ViT models (i.e., 0.825, 0.825, 0.826, 0.964, 0.825, 0.825, and 0.956 for F1-score, accuracy, balanced accuracy, AUC, precision, recall, specificity, respectively). The results of the proposed ViT model were quite encouraging to be applied in real medical environment for assisting physicians to make accurate, personalized, and timely decisions.

Many companies have transformed their telephone systems into Voice over IP (VoIP) systems. Although implementation is simple, VoIP is vulnerable to different types of attacks. The Session Initiation Protocol (SIP) is a widely used protocol for handling VoIP signaling functions. SIP is unprotected against attacks because it is a text-based protocol and lacks defense against the growing security threats. The Distributed Denial of Service (DDoS) attack is a harmful attack, because it drains resources, and prevents legitimate users from using the available services. In this paper, we formulate detection of DDoS attacks as a classification problem and propose an approach using token embedding to enhance extracted features from SIP messages. We discuss a deep learning model based on Recurrent Neural Networks (RNNs) developed to detect DDoS attacks with low and high-rate intensity. For validation, a balanced real traffic dataset was built containing three attack scenarios with different attack durations and intensities. Experiments show that the system has a high detection accuracy and low detection time. The detection accuracy was higher for low-rate attacks than that of traditional machine learning.

Mental health disorders (MHDs) have significant medical and financial impacts on patients and society. Despite the potential opportunities for artificial intelligence (AI) in the mental health field, there are no noticeable roles of these systems in real medical environments. The main reason for these limitations is the lack of trust by domain experts in the decisions of AI-based systems. Recently, trustworthy AI (TAI) guidelines have been proposed to support the building of responsible AI (RAI) systems that are robust, fair, and transparent. This review aims to investigate the literature of TAI for machine learning (ML) and deep learning (DL) architectures in the MHD domain. To the best of our knowledge, this is the first study that analyzes the literature of trustworthiness of ML and DL models in the MHD domain. The review identifies the advances in the literature of RAI models in the MHD domain and investigates how this is related to the current limitations of the applicability of these models in real medical environments. We discover that the current literature on AI-based models in MHD has severe limitations compared to other domains regarding TAI standards and implementations. We discuss these limitations and suggest possible future research directions that could handle these challenges.

In the software development process, more than one developer may work on developing the same program and bugs in the program may be fixed by a different developer; therefore, understanding the source code is an important issue. Pseudocode plays an important role in solving this problem, as it helps the developer to understand the source code. Recently, transformer-based pre-trained models achieved remarkable results in machine translation, which is similar to pseudocode generation. In this paper, we propose a novel automatic pseudocode generation from the source code based on a pre-trained Bidirectional and Auto-Regressive Transformer (BART) model. We fine-tuned two pre-trained BART models (i.e., large and base) using a dataset containing source code and its equivalent pseudocode. In addition, two benchmark datasets (i.e., Django and SPoC) were used to evaluate the proposed model. The proposed model based on the BART large model outperforms other state-of-the-art models in terms of BLEU measurement by 15% and 27% for Django and SPoC datasets, respectively.

The comprehension of source code is very difficult, especially if the programmer is not familiar with the programming language. Pseudocode explains and describes code contents that are based on the semantic analysis and understanding of the source code. In this paper, a novel retrieval-based transformer pseudocode generation model is proposed. The proposed model adopts different retrieval similarity methods and neural machine translation to generate pseudocode. The proposed model handles words of low frequency and words that do not exist in the training dataset. It consists of three steps. First, we retrieve the sentences that are similar to the input sentence using different similarity methods. Second, pass the source code retrieved (input retrieved) to the deep learning model based on the transformer to generate the pseudocode retrieved. Third, the replacement process is performed to obtain the target pseudo code. The proposed model is evaluated using Django and SPoC datasets. The experiments show promising performance results compared to other language models of machine translation. It reaches 61.96 and 50.28 in terms of BLEU performance measures for Django and SPoC, respectively.

Voice over IP (VoIP) services hold promise because of their offered features and low cost. Most VoIP networks depend on the Session Initiation Protocol (SIP) to handle signaling functions. The SIP is a text-based protocol that is vulnerable to many attacks. Denial of Service (DoS) and distributed denial of service (DDoS) attacks are the most harmful types of attacks, because they drain VoIP resources and render SIP service unavailable to legitimate users. In this paper, we present recently introduced approaches to detect DoS and DDoS attacks, and classify them based on various factors. We then analyze these approaches according to various characteristics; furthermore, we investigate the main strengths and weaknesses of these approaches. Finally, we provide some remarks for enhancing the surveyed approaches and highlight directions for future research to build effective detection solutions.