Search

Brain tumors pose a significant threat, especially when not detected early. The Inception v3 machine learning model has found extensive applications in computer vision and related fields. This study aims to develop a robust transfer learning model for classification, adaptable to various data modalities through neural networks. However, the training process for these neural networks is complex, being both demanding and computationally intensive. To tackle this challenge, we introduce an innovative training approach for Inception v3 referred to as ‘PSO-INCEPT’ (Particle Swarm Optimization-based Inception v3 training). In this method, the weight vectors for each Inception v3 model are analogized to particle positions in a phase space. The PSO cooperates with the ADAM optimizer in achieving the purpose of training with the best performance and generalization. This research is composed of two main parts, the first stage is being performed by the model independently using the ADAM optimizer. In the following stage, PSO-INCEPT models share the latest weight vectors or particle coordinates and loss function approximations via training. The optimization function then uses them to improve the validation accuracy. The effectiveness of PSO-INCEPT was evaluated through experiments that were conducted on Kaggle datasets that provide a common base ground by having four distinct classes. Experimental studies have proven the extraordinary ability of the proposed model by providing 99.33% validation accuracy and 99.95% training accuracy which shows exceptional performance.

A growing kind of communication, the Internet of Things (IoT), links all network-capable devices worldwide. The most recent phase of the Internet of Things (IoT), known as the Internet of Medical Things (IoMT), is one that is rapidly catching researchers’ interest. The centralized storage system houses the substantial amount of medical data created by IoMT. However, centralizing sensitive patient data creates a single point of failure and raises privacy and security issues. To assure honesty, reliability, and safety while determining an entity’s trustworthiness, trust information is required to be safely transmitted and preserved. We propose a safe trust management solution based on blockchain technology. The proposed solution gathers node trust data using both time- and event-driven methods that are used to compute the trust of a node based on the threshold value. We also confirm the validity of data by varying the threshold value during transmission and authentication and employing a membership mechanism for authentication. The trust scores for each node are securely stored in an array, and the Maximum subarray (Kadane) Algorithm is used to compute the threshold value. The blockchain network receives and stores the IoMT score in the trusted list. Due to its distributed structure, capacity to maintain safeguards, and resiliency against a variety of threats with low overhead. Our approach is viable, deployable, and appropriate and also includes safety features like tamper-proofing, attack resistance, dependability, and low capabilities for IoT in smart hospitals.

Parkinson’s disease affects bodily functions and there is a growing need for advanced solutions to offer therapeutic advice to patients. A framework using artificial intelligence and machine learning techniques has been proposed to address this. The proposed system employs a combination of RGB, inertial, and depth sensors data. The inertial signals have been filtered using a notch filter to obtain the optimal wearable sensor data by examining the upper and lower cutoff frequencies. Multiple features have been calculated, including mel frequency cepstral coefficients (MFCC), statistical features and Gaussian mixture model (GMM) features. On the other hand, silhouettes have been extracted from RGB and depth images, and four crucial parameters have been employed to gauge the level of accuracy with which patients with neurological disorders performed their activities, including the angle formation between the hands and lower/upper half of the body and center of the body, and the angle between the hands of the silhouette. The resulting features have been fused and classified using principle component analysis and a reweighted genetic algorithm. Evaluation using cross-validation on mRI (multi-modal 3d human pose estimation dataset using mmwave, RGB-D, and inertial sensors) and MHEALTH (mobile health) datasets showed a recognition accuracy rate of 97.29% and a 97.94%. The study highlights the need for more datasets to address challenges in rehabilitation using human activity recognition with multi-modal sensors.

The Industrial Internet of Things (IIoT) comprises a variety of systems, smart devices, and an extensive range of communication protocols. Hence, these systems face susceptibility to privacy and security challenges, making them prime targets for malicious attacks that can result in harm to the overall system. Privacy breach issues are a notable concern within the realm of IIoT. Various intrusion detection systems based on machine learning (ML) and deep learning (DL) have been introduced to detect malicious activities within these networks and identify attacks. However, traditional ML and DL models encounter significant hurdles when faced with highly imbalanced training data and repetitive patterns within network datasets, hampering their performance in distinguishing between various classes of attacks. To overcome the challenges inherent in existing systems, this paper presents a self-attention-based deep convolutional neural network (SA-DCNN) model designed for monitoring the IIoT networks and detecting malicious activities. The SA mechanism computes the significance value for each input feature, and the DCNN processes these parameters to detect IIoT network behavior. Additionally, a two-step cleaning method has been implemented to eliminate redundancy within the training data, considering both intra-class and cross-class samples. Furthermore, to tackle the issue of underfitting, we have employed a mutual information-based feature filtering method. This method ranks all the features in descending order based on their mutual information and subsequently removes the features with negative impact from the dataset. The performance of the SA-DCNN model is assessed using IoTID20 and Edge-IIoTset datasets. Moreover, the proposed study is demonstrated through a comprehensive comparison with other ML and DL models, as well as against relevant studies, showcasing the superior performance and efficacy of the proposed model.

Cervical cancer, the second most prevalent cancer among women worldwide, is primarily attributed to the human papillomavirus (HPV). Despite advances in healthcare, it remains a significant cause of mortality among women across diverse regions, surpassing other hereditary cancers. Early detection is pivotal, as survival rates exceed 90% when the disease is identified in its early stages. In response to this critical need, we introduce WFC2DS (Web Framework for Cervical Cancer Detection System), a novel expert web system specifically designed to revolutionize cervical cancer diagnosis. WFC2DS integrates a sophisticated ensemble of machine learning classification algorithms, including Artificial Neural Network (ANN), AdaBoost, K-Nearest Neighbor (KNN), Random Forest Classifier (RFC), Support Vector Machine (SVM), and Decision Tree (DT). This ensemble approach enables a comprehensive analysis of a large dataset comprising information from 858 patients with 36 attributes, with the primary objective being the early detection of cervical cancer, using the last attribute, Biopsy, as the target variable. Our evaluation criteria encompass accuracy, specificity, sensitivity, and the F1 score. Among the algorithms, RFC and DT emerge as the most promising, demonstrating exceptional performance with an accuracy of 98.1% and an F1 score of 0.98. AdaBoost shows an accuracy of 97.4% and an F1 score of 0.98, ANN attains an accuracy of 97.7% and an F1 score of 0.96, SVM achieves an accuracy of 96.2% and an F1 score of 0.96, and KNN reaches an accuracy of 90.6% with an F1 score of 0.91. This research significantly contributes to reducing the global burden of cervical cancer, emphasizing transformative advancements in women’s healthcare. WFC2DS, with its cutting-edge machine learning techniques, not only improves the accuracy of cervical cancer diagnosis but also enhances the overall healthcare landscape for women worldwide.

The rise of Internet of Things (IoT) has led to increased security risks, particularly from botnet attacks that exploit IoT device vulnerabilities. This situation necessitates effective Intrusion Detection Systems (IDS), that are accurate, lightweight, and fast (having less inference time), designed particularly to detect botnet attacks in resource constrained IoT devices. This paper proposes SkipGateNet, a novel deep learning model designed for detecting Mirai and Bashlite botnet attacks in resource constrained IoT and fog computing environments. SkipGateNet is a lightweight, fast model combining 1D-Convolutional Neural Networks (CNN) and Long Short-Term Memory (LSTM) layers. The novelty of this model lies in the integration of ‘Learnable Skip Connections’. These connections feature gating mechanisms that enhance detection by focusing on relevant features and ignoring irrelevant ones. They add adaptability to the architecture, performing feature selection and propagating only essential features to deeper layers. Tested on the N-BaIoT dataset, SkipGateNet efficiently detects ten types of botnet attacks, with a remarkable test accuracy of 99.91%. It is also compact (2596.87 KB) and demonstrates a quick inference time of 8.0 milliseconds, suitable for real-time implementation in resource-limited settings. While evaluating its performance, parameters like precision, recall, accuracy, and F1 score were considered, along with statistical reliability measures like Cohen’s Kappa Coefficient and Matthews Correlation Coefficient. These highlight its reliability and effectiveness in IoT security challenges. The paper also compares SkipGateNet to existing models and four other deep learning architectures, including two sequential CNN architectures, a simple CNN+LSTM architecture, and a CNN+LSTM with standard skip connections. SkipGateNet surpasses all in accuracy and inference time, demonstrating its superiority in addressing IoT security issues.

The emergence of two new technologies, namely software defined networking (SDN) and 5G networks, has greatly changed the development of network functions and network topologies. These two technologies provide cost benefits for mobile operators, a more flexible and scalable network, and a shorter time to market for new services and applications. Scalability and effectiveness are increased when 5G and SDN are used together. SDN increases the reliability of the 5G network by separating the control plane from the data plane. Incorrect load balancing, a lack of knowledge of network traffic, and other issues make it difficult to provide Quality of Service (QoS) with SDN. This research proposes a unique load-balancing method to resolve these concerns using Hierarchical Agglomerative Clustering (HAC) and Back Propagation Neural Network (BPNN) algorithms. The proposed method segments the network services into several groups after normalizing data requirements. It consists of two phases: in first phase, there is clustering of bandwidth for different services (e.g., social media, automated homes, and automated cars) inside the network for different data requirements. The agglomerative hierarchical clustering (single link technique) is implemented to make the clusters of bandwidth inside the SDN work based on minimum distance. After clustering, we allotted bandwidths to the respective clusters. In the second phase, the BPNN technique trains the network to choose the optimal path and check the error faults. The proposed approach evaluates the network delay, packet loss, throughput, latency rate, and bandwidth usage to evaluate the performance with Multiple Regression-based Searching (MRBS) and Software-defined Sensor Network Load Balancing (SDSNLB) algorithms. The experimental results of the proposed model are promising as the performance increased by 15%, 23%, 27%, 21%, and 30%, respectively, compared with existing approaches. In addition we compared the computational time complexity with increased the no of nodes and services, when the rate of nodes and services is varying, the proposed solution’s efficiency remains constant.

There are many problems that procedural algorithms can solve efficiently. However, these algorithms are sometimes too slow to abide by the time available for performing the solution; other times, it is impossible to get a solution using procedural algorithms. A heuristic method is a practical approach that can reach an approximation of an efficient solution where the optimum is not guaranteed. Heuristic techniques are applied in many real-world problems, including crowd management; using heuristic-based models helped to comprehend crowd behavior better and increase simulation reliability. This paper reviews many heuristic-related articles to gather the aspects of the topic in one place and clear the fuzziness to make it easy to comprehend. The paper covers some of the previous works with similar approaches and presents state-of-the-art heuristic solutions for real-world problems. These techniques are discussed under three classifications: simple heuristics, meta-heuristics, and hyper-heuristics. Most importantly, the paper explores the heuristic role in crowd field problems concluding that heuristics are primarily applied in modeling when it comes to the Crowd field. It investigates different heuristics for crowd management. The main intent of this review is to establish a comprehensive understanding of heuristics-related operations in the crowd management field. Moreover, it aims to support other researchers’ future work and fill research gaps by highlighting the absence of crowd problems from heuristics literature and the limitations of each heuristics approach.

Recently, database users have begun to use cloud database services to outsource their databases. The reason for this is the high computation speed and the huge storage capacity that cloud owners provide at low prices. However, despite the attractiveness of the cloud computing environment to database users, privacy issues remain a cause for concern for database owners since data access is out of their control. Encryption is the only way of assuaging users’ fears surrounding data privacy, but executing Structured Query Language (SQL) queries over encrypted data is a challenging task, especially if the data are encrypted by a randomized encryption algorithm. Many researchers have addressed the privacy issues by encrypting the data using deterministic, onion layer, or homomorphic encryption. Nevertheless, even with these systems, the encrypted data can still be subjected to attack. In this research, we first propose an indexing scheme to encode the original table’s tuples into bit vectors (BVs) prior to the encryption. The resulting index is then used to narrow the range of retrieved encrypted records from the cloud to a small set of records that are candidates for the user’s query. Based on the indexing scheme, we then design a system to execute SQL queries over the encrypted data. The data are encrypted by a single randomized encryption algorithm, namely the Advanced Encryption Standard-Cipher-Block Chaining (AES-CBC). In the proposed scheme, we store the index values (BVs) at user’s side, and we extend the system to support most of relational algebra operators, such as select, join, etc. Implementation and evaluation of the proposed system reveals that it is practical and efficient at reducing both the computation and space overhead when compared with state-of-the-art systems like CryptDB.

Crowd management is a flourishing, active research area and must be given attention due to the potential losses, disasters, and accidents that could occur if it were neglected. For the last decade, the crowd management field has witnessed significant advancements; however, more investigative work is still needed. The integration of different crowd detection and monitoring techniques can enhance the control and the performance compared to those of more limited stand-alone techniques. Crowd management encompasses an entire process, from the monitoring stage through the decision support system stage. This sector involves accessing and interpreting information sources, predicting crowd behavior, and deciding on the use of a range of possible interventions based on context. This paper shows a fresh conclusive review of the concept of the crowd, discussing it from several perspectives in light of its defining characteristics, its risks, and tragedies, which may occur due to challenges faced during crowd management, where these conclusions are based on a massive number of scholarly articles that were newly published. Besides, a systematic discussion is shown concerning the steps of managing a crowd, including crowd detection, in which several new methods are reviewed, followed by illustrating both direct and indirect approaches to crowd monitoring and tracking monitoring. The primary purpose of this review is to establish a comprehensive understanding of crowd-related processes. Moreover, it aims to find research gaps to overcome the limitations of using stand-alone techniques in each process and provide support to other researchers' future work.