Mohammed Ali

Accurate 3D object detection is vital in autonomous driving. Single-modal detectors, using either camera or LiDAR, struggle with issues like limited depth perception or difficulty in distinguishing semantically similar objects. While multimodal approaches aim to address these limitations by combining LiDAR and camera data, they often face complexities in integrating sparse and uneven point cloud distributions, resulting in inefficient feature fusion. To tackle these challenges, we propose CLS-3D (Content-wise LiDAR-Camera Fusion and Slot Reweighting Transformer for 3D Object Detection). This novel framework fuses LiDAR and camera features using a single multi-modal backbone and augments them with the semantic probabilities obtained from the image stream. Our method captures local and global spatial relationships through a slot reweighting mechanism and incorporates I3C-IoU loss for precise box regression. The semantically augmented features, via a single multi-modal backbone, are embedded using a content-based transformer and processed through a slot-wise auto-encoder structure with channel-wise positional embeddings and a feed-forward MLP network. Our model improves temporal consistency and detection accuracy by dynamically adjusting feature relevance through slot-wise reweighting. We further define a I3C-IoU metric, considering centre, overlap, and scale for enhanced box regression accuracy. This mechanism allows the model to focus on significant temporal information, enhancing its ability to learn complex sequences and improving the overall performance of 3D object detection, especially in challenging scenarios such as occlusion and long-range detection. Extensive experiments on the KITTI and nuScenes benchmark demonstrate that CLS-3D achieves state-of-the-art performance, with 89.52% 3D mAP and 94.08% BEV mAP, outperforming existing methods.

Meeting the energy needs of a growing population is of paramount importance in today’s society. The use of renewable energy sources, especially solar energy, can help reduce greenhouse gas emissions from traditional sources. Solar irradiance, which depends on several factors such as the availability of sunlight, cloud cover index, latitude, orientation and tilt of solar panels, and technical factors, plays a crucial role in the use of solar energy. Predicting solar irradiance is therefore essential for enhancing the reliability and efficiency of solar energy systems. Artificial intelligence (AI) and machine learning (ML) models are increasingly being used to predict solar irradiance forecasting. These models can learn from historical weather data and identify complex patterns to predict future solar irradiance with high accuracy. Techniques such as regression, neural networks and ensemble methods are frequently used and enable more efficient planning and operation of solar energy systems. This study provides a comprehensive analysis of the existing state-of-the-art models for solar irradiance forecasting. The study evaluates a variety of forecasting models, including machine learning, numerical weather prediction and hybrid approaches for assessing their accuracy, strengths and weaknesses. As an outcome of this study, potential future improvements in the prediction of solar irradiance are highlighted and the importance of interdisciplinary collaboration and emerging technologies is emphasized. To ensure a sustainable and resilient energy future, it is crucial to continue efforts to better integrate solar energy into mainstream electricity systems. This work can serve as a fundamental analysis for future researchers to identify the most appropriate approaches for medium and long-term solar irradiance forecasts.

Improving software quality by proactively detecting potential defects during development is a major goal of software engineering. Software defect prediction plays a central role in achieving this goal. The power of data analytics and machine learning allows us to focus our efforts where they are needed most. A key factor in the success of software fault prediction is selecting relevant features and reducing data dimensionality. Feature selection methods contribute by filtering out the most critical attributes from a plethora of potential features. These methods have the potential to significantly improve the accuracy and efficiency of fault prediction models. However, the field of feature selection in the context of software fault prediction is vast and constantly evolving, with a variety of techniques and tools available. Based on these considerations, our systematic literature review conducts a comprehensive investigation of feature selection methods used in the context of software fault prediction. The research uses a refined search strategy involving four reputable digital libraries, including IEEE Explore, Science Direct, ACM Digital Library, and Springer Link, to provide a comprehensive and exhaustive review through a rigorous analysis of 49 selected primary studies from 2014. The results highlight several important issues. First, there is a prevalence of filtering and hybrid feature selection methods. Second, single classifiers such as Naïve Bayes, Support Vector Machine, and Decision Tree, as well as ensemble classifiers such as Random Forest, Bagging, and AdaBoost are commonly used. Third, evaluation metrics such as area under the curve, accuracy, and F-measure are commonly used for performance evaluation. Finally, there is a clear preference for tools such as WEKA, MATLAB, and Python. By providing insights into current trends and practices in the field, this study offers valuable guidance to researchers and practitioners to make informed decisions to improve software fault prediction models and contribute to the overall improvement of software quality.

Recent advances in information and communication technology have greatly increased the popularity of online education. The COVID-19 pandemic has further accelerated the adoption of online education due to the need for educational institutions to adopt online learning models. On the other hand, the online education system faces several challenges, such as lack of usability to overcome access barriers and scalability issues. These problems stand in the way of widespread adoption of the online education system. The usability of online education websites is one of the fundamental challenges because it directly affects students’ learning experience. This study focuses on the challenges related to usability of online education websites. If an online education website has usability issues, students may have difficulty using the learning materials provided. This makes it difficult to achieve the required level of service, which can lead to a poor reputation for the website. Therefore, in order to develop a well-functioning online education website, it is necessary to ensure that all usability requirements (URs) are met efficiently. Considering the different backgrounds, knowledge, and skills of students, they may have different preferences regarding usability requirements. Therefore, prioritization of URs is necessary to focus on the most preferred requirements. In this study, we apply the fuzzy analytic hierarchy process (FAHP) to analyze the usability requirements in the requirements analysis phase. We consider the online education website of Virtual University of Pakistan, a leading online university in the country with more than 0.1 million active students. The result of this work is a flexible approach to identify precise usability requirements for developing a functionally sound website with an optimized and easy-to-use user interface that will ultimately improve the quality of education.

The Internet of Things (IoT) and wireless sensor networks (WSNs) have evolved rapidly due to technological breakthroughs. WSNs generate high traffic due to the growing number of sensor nodes. Congestion is one of several problems caused by the huge amount of data in WSNs. When wireless network resources are limited and IoT devices require more and more resources, congestion occurs in extremely dense WSN-based IoT networks. Reduced throughput, reduced network capacity, and reduced energy efficiency within WSNs are all effects of congestion. These consequences eventually lead to network outages due to underutilized network resources, increased network operating costs, and significantly degraded quality of service (QoS). Therefore, it is critical to deal with congestion in WSN-based IoT networks. Researchers have developed a number of approaches to address this problem, with new solutions based on artificial intelligence (AI) standing out. This research examines how new AI-based algorithms contribute to congestion mitigation in WSN-based IoT networks and the various congestion mitigation strategies that have helped reduce congestion. This study also highlights the limitations of AI-based solutions, including where and why they are used in WSNs, and a comparative study of the current literature that makes this study novel. The study concludes with a discussion of its significance and potential future study topics. The topic of congestion reduction in ultra-dense WSN-based IoT networks, as well as the current state of the art and emerging future solutions, demonstrates their significant expertise in reducing WSN congestion. These solutions contribute to network optimization, throughput enhancement, quality of service improvement, network capacity expansion, and overall WSN efficiency improvement.

The limited resources and enormous amounts of data generated by multimedia sensors require efficient strategies to extend network lifetime while taking into account quality-of-service requirements such as reliability and delay. In contrast, limited battery resources require new techniques to balance energy consumption and multimedia application requirements in wireless multimedia sensor networks (WMSNs). These requirements are very critical, especially for network stability and performance. In this paper, an energy-efficient mechanism based on the M/D/1/B queueing model is proposed. According to the packets in the queue and the waiting time, the nodes decide their activation time, so the nodes wake up for a while to transmit the data in the queue and then go to sleep mode. The simulation results of the proposed algorithm show that the proposed mechanism achieves optimal values to reduce energy consumption while meeting the quality-of-service requirements under different conditions.

The rapid advancement of networking, computing, sensing, and control systems has introduced a wide range of cyber threats, including those from new devices deployed during the development of scenarios. With recent advancements in automobiles, medical devices, smart industrial systems, and other technologies, system failures resulting from external attacks or internal process malfunctions are increasingly common. Restoring the system’s stable state requires autonomous intervention through the self-healing process to maintain service quality. This paper, therefore, aims to analyse state of the art and identify where self-healing using machine learning can be applied to cyber–physical systems to enhance security and prevent failures within the system. The paper describes three key components of self-healing functionality in computer systems: anomaly detection, fault alert, and fault auto-remediation. The significance of these components is that self-healing functionality cannot be practical without considering all three. Understanding the self-healing theories that form the guiding principles for implementing these functionalities with real-life implications is crucial. There are strong indications that self-healing functionality in the cyber–physical system is an emerging area of research that holds great promise for the future of computing technology. It has the potential to provide seamless self-organising and self-restoration functionality to cyber–physical systems, leading to increased security of systems and improved user experience. For instance, a functional self-healing system implemented on a power grid will react autonomously when a threat or fault occurs, without requiring human intervention to restore power to communities and preserve critical services after power outages or defects. This paper presents the existing vulnerabilities, threats, and challenges and critically analyses the current self-healing theories and methods that use machine learning for cyber–physical systems.

Human action recognition is a constantly evolving field that is driven by numerous applications. In recent years, significant progress has been made in this area due to the development of advanced representation learning techniques. Despite this progress, human action recognition still poses significant challenges, particularly due to the unpredictable variations in the visual appearance of an image sequence. To address these challenges, we propose the fine-tuned temporal dense sampling with 1D convolutional neural network (FTDS-1DConvNet). Our method involves the use of temporal segmentation and temporal dense sampling, which help to capture the most important features of a human action video. First, the human action video is partitioned into segments through temporal segmentation. Each segment is then processed through a fine-tuned Inception-ResNet-V2 model, where max pooling is performed along the temporal axis to encode the most significant features as a fixed-length representation. This representation is then fed into a 1DConvNet for further representation learning and classification. The experiments on UCF101 and HMDB51 demonstrate that the proposed FTDS-1DConvNet outperforms the state-of-the-art methods, with a classification accuracy of 88.43% on the UCF101 dataset and 56.23% on the HMDB51 dataset.

Autonomous vehicles have become a topic of interest in recent times due to the rapid advancement of automobile and computer vision technology. The ability of autonomous vehicles to drive safely and efficiently relies heavily on their ability to accurately recognize traffic signs. This makes traffic sign recognition a critical component of autonomous driving systems. To address this challenge, researchers have been exploring various approaches to traffic sign recognition, including machine learning and deep learning. Despite these efforts, the variability of traffic signs across different geographical regions, complex background scenes, and changes in illumination still poses significant challenges to the development of reliable traffic sign recognition systems. This paper provides a comprehensive overview of the latest advancements in the field of traffic sign recognition, covering various key areas, including preprocessing techniques, feature extraction methods, classification techniques, datasets, and performance evaluation. The paper also delves into the commonly used traffic sign recognition datasets and their associated challenges. Additionally, this paper sheds light on the limitations and future research prospects of traffic sign recognition.