Recent advances in Convolutional Neural Networks (CNNs) have obtained promising results in difficult deep learning tasks. However, the success of a CNN depends on finding an architecture to fit a given problem. A hand-crafted architecture is a challenging, time-consuming process that requires expert knowledge and effort, due to a large number of architectural design choices. In this article, we present an efficient framework that automatically designs a high-performing CNN architecture for a given problem. In this framework, we introduce a new optimization objective function that combines the error rate and the information learnt by a set of feature maps using deconvolutional networks (deconvnet). The new objective function allows the hyperparameters of the CNN architecture to be optimized in a way that enhances the performance by guiding the CNN through better visualization of learnt features via deconvnet. The actual optimization of the objective function is carried out via the Nelder-Mead Method (NMM). Further, our new objective function results in much faster convergence towards a better architecture. The proposed framework has the ability to explore a CNN architecture’s numerous design choices in an efficient way and also allows effective, distributed execution and synchronization via web services. Empirically, we demonstrate that the CNN architecture designed with our approach outperforms several existing approaches in terms of its error rate. Our results are also competitive with state-of-the-art results on the MNIST dataset and perform reasonably against the state-of-the-art results on CIFAR-10 and CIFAR-100 datasets. Our approach has a significant role in increasing the depth, reducing the size of strides, and constraining some convolutional layers not followed by pooling layers in order to find a CNN architecture that produces a high recognition performance.

Although deep learning algorithms have achieved significant progress in a variety of domains, they require costly annotations on huge datasets. Self-supervised learning (SSL) using unlabeled data has emerged as an alternative, as it eliminates manual annotation. To do this, SSL constructs feature representations using pretext tasks that operate without manual annotation, which allows models trained in these tasks to extract useful latent representations that later improve downstream tasks such as object classification and detection. The early methods of SSL are based on auxiliary pretext tasks as a way to learn representations using pseudo-labels, or labels that were created automatically based on the dataset’s attributes. Furthermore, contrastive learning has also performed well in learning representations via SSL. To succeed, it pushes positive samples closer together, and negative ones further apart, in the latent space. This paper provides a comprehensive literature review of the top-performing SSL methods using auxiliary pretext and contrastive learning techniques. It details the motivation for this research, a general pipeline of SSL, the terminologies of the field, and provides an examination of pretext tasks and self-supervised methods. It also examines how self-supervised methods compare to supervised ones, and then discusses both further considerations and ongoing challenges faced by SSL.

Currently, treating sign language issues and producing high quality solutions has attracted researchers and practitioners’ attention due to the considerable prevalence of hearing disabilities around the world. The literature shows that Arabic Sign Language (ArSL) is one of the most popular sign languages due to its rate of use. ArSL is categorized into two groups: The first group is ArSL, where words are represented by signs, i.e., pictures. The second group is ArSl alphabetic (ArSLA), where each Arabic letter is represented by a sign. This paper introduces a real time ArSLA recognition model using deep learning architecture. As a methodology, the proceeding steps were followed. First, a trusted scientific ArSLA dataset was located. Second, the best deep learning architectures were chosen by investigating related works. Third, an experiment was conducted to test the previously selected deep learning architectures. Fourth, the deep learning architecture was selected based on extracted results. Finally, a real time recognition system was developed. The results of the experiment show that the AlexNet architecture is the best due to its high accuracy rate. The model was developed based on AlexNet architecture and successfully tested at real time with a 94.81% accuracy rate.

Robot navigation in indoor environments has become an essential task for several applications, including situations in which a mobile robot needs to travel independently to a certain location safely and using the shortest path possible. However, indoor robot navigation faces challenges, such as obstacles and a dynamic environment. This paper addresses the problem of social robot navigation in dynamic indoor environments, through developing an efficient SLAM-based localization and navigation system for service robots using the Pepper robot platform. In addition, this paper discusses the issue of developing this system in a way that allows the robot to navigate freely in complex indoor environments and efficiently interact with humans. The developed Pepper-based navigation system has been validated using the Robot Operating System (ROS), an efficient robot platform architecture, in two different indoor environments. The obtained results show an efficient navigation system with an average localization error of 0.51 m and a user acceptability level of 86.1%.

Diabetes is a growing global health concern, affecting millions and leading to severe complications if not properly managed. The primary challenge in diabetes management is maintaining blood glucose levels (BGLs) within a safe range to prevent complications such as renal failure, cardiovascular disease, and neuropathy. Traditional methods, such as finger-prick testing, often result in low patient adherence due to discomfort, invasiveness, and inconvenience. Consequently, there is an increasing need for non-invasive techniques that provide accurate BGL measurements. Photoplethysmography (PPG), a photosensitive method that detects blood volume variations, has shown promise for non-invasive glucose monitoring. Deep neural networks (DNNs) applied to PPG signals can predict BGLs with high accuracy. However, training DNN models requires large and diverse datasets, which are typically distributed across multiple healthcare institutions. Privacy concerns and regulatory restrictions further limit data sharing, making conventional centralized machine learning (ML) approaches less effective. To address these challenges, this study proposes a federated learning (FL)-based solution that enables multiple healthcare organizations to collaboratively train a global model without sharing raw patient data, thereby enhancing model performance while ensuring data privacy and security. In the data preprocessing stage, continuous wavelet transform (CWT) is applied to smooth PPG signals and remove baseline drift. Adaptive cycle-based segmentation (ACBS) is then used for signal segmentation, followed by particle swarm optimization (PSO) for feature selection, optimizing classification accuracy. The proposed system was evaluated on diverse datasets, including VitalDB and MUST, under various conditions with data collected during surgery and anesthesia. The model achieved a root mean square error (RMSE) of 19.1 mg/dL, demonstrating superior predictive accuracy. Clarke error grid analysis (CEGA) confirmed the model’s clinical reliability, with 99.31% of predictions falling within clinically acceptable limits. The FL-based approach outperformed conventional deep learning models, making it a promising method for non-invasive, privacy-preserving glucose monitoring.

Football is one of the most-watched sports, but analyzing players’ performance is currently difficult and labor intensive. Performance analysis is done manually, which means that someone must watch video recordings and then log each player’s performance. This includes the number of passes and shots taken by each player, the location of the action, and whether or not the play had a successful outcome. Due to the time-consuming nature of manual analyses, interest in automatic analysis tools is high despite the many interdependent phases involved, such as pitch segmentation, player and ball detection, assigning players to their teams, identifying individual players, activity recognition, etc. This paper proposes a system for developing an automatic video analysis tool for sports. The proposed system is the first to integrate multiple phases, such as segmenting the field, detecting the players and the ball, assigning players to their teams, and identifying players’ jersey numbers. In team assignment, this research employed unsupervised learning based on convolutional autoencoders (CAEs) to learn discriminative latent representations and minimize the latent embedding distance between the players on the same team while simultaneously maximizing the distance between those on opposing teams. This paper also created a highly accurate approach for the real-time detection of the ball. Furthermore, it also addressed the lack of jersey number datasets by creating a new dataset with more than 6,500 images for numbers ranging from 0 to 99. Since achieving a high performance in deep learning requires a large training set, and the collected dataset was not enough, this research utilized transfer learning (TL) to first pretrain the jersey number detection model on another large dataset and then fine-tune it on the target dataset to increase the accuracy. To test the proposed system, this paper presents a comprehensive evaluation of its individual stages as well as of the system as a whole. All codes, datasets, and experiments are available on GitHub ().