Due to increasing maritime activities, the number of Maritime Internet-of-things (MIoT) devices requiring real-time marine data processing is growing exponentially. To offload maritime tasks and address the limited computational capabilities of heterogeneous MIoT devices, edge and cloud computing networks are employed. However, these networks introduce several challenges, including increased energy consumption and service latency within the complex marine network environment. Current state-of-the-art solutions address these issues by focusing exclusively on real-time offloading data, neglecting the relationship with past offloading tasks. In this work, we develop an optimization framework, named VESBELT, for offloading tasks from Vessel users to nearby Edge Servers or the cloud server, aiming to reduce Energy consumption and service Latency Trade-off through a multi-objective linear programming problem. However, finding optimal solutions from this formulation is considered an NP-hard problem. To address this, we introduce VESBELT-ECNN, VESBELT-EANN, and VESBELT-ELSTM systems that leverage ensemble convolutional, artificial neural networks and Long-short-term memory, respectively, to achieve solutions in polynomial time. The developed ensemble models integrate multiple combinations of deep learning models and exploit the pre-trained models to provide real-time solutions with better prediction accuracy. The experimental findings, obtained using Python programming version 3.10.2, indicate that the proposed VESBELT-ECNN, VESBELT-EANN, and VESBELT-ELSTM systems outperform existing approaches in terms of user Quality of Experience (QoE) in the timeliness domain and energy savings.

Image classification is a fundamental task in computer vision with numerous applications. Even though, the rapid development of Convolutional Neural Networks (CNNs) has revolutionized the field of image classification, there is a constant pursuit for further improvement. Existing CNN models typically rely solely on downscaling techniques for pre-processing, limiting their potential for extracting intricate image details and effectively handling varying image resolutions. In this paper, we propose an approach that incorporates both upscaling and down-scaling techniques to enhance the performance of CNN models for image classification. Our method leverages the benefits of both upscaling and downscaling operations to capture fine-grained details and extract high-level features, respectively. Through extensive experiments on benchmark datasets, namely CIFAR-10 and CIFAR-100 dataset, we demonstrate the superiority of our proposed approach over existing methods in terms of accuracy and robustness. For instance, our proposed model, named Super-Res VGG has received 85.57% testing accuracy whereas VGG16, GoogLeNet (Inception), and ResNet50 have received 84.40%, 73.93%, and 65.29% accuracy on CIFAR-10 dataset, respectively.

Serverless edge computing is increasingly adopted in smart cities and industrial automation applications, which leverages cloud computing for scalable resources and uses Function as a Service (FaaS) for efficient load balancing, pay-as-you-go execution, and third-party management of provisioning and auto-scaling. However, task scheduling in serverless environments is challenging due to service latency, provider costs, and cold start issues caused by dynamic workloads and resource availability. While existing literature has focused on task scheduling, it often overlooks the joint minimization of service delay and provider costs, as well as long-term workloads, and energy use. In this work, we propose an optimization framework using mixed integer linear programming (MILP) to jointly minimize task execution delay and provider costs, namely DECASE. Given the NP-hard nature of the optimization for large networks, we developed a metaheuristic Artificial Bee Colony (ABC) algorithm to provide near-optimal task scheduling and resource allocation within polynomial time. The developed DECASE system significantly reduces delays and serverless resource provider costs by up to 20% and 25% compared to existing methods.

To meet the demand of the world's largest population, smart manufacturing has accelerated the adoption of smart factories-where autonomous and cooperative instruments across all levels of production and logistics networks are integrated through a Cyber-Physical Production System (CPPS). However, these networks are comprised of various heterogeneous devices with varying computational power and memory capabilities. As a result, many secure communication protocols - that demand considerably high computational power and memory - can not be verbatim employed on these networks, and thereby, leaving them more vulnerable to security threats and attacks over conventional networks. These threats can largely be tackled by employing a Trust Management Model (TMM) by exploiting the behavioural patterns of nodes to identify their trust class. In this context, ML-based models are best suited due to their ability to capture hidden patterns in data, learning and improving the pattern detection accuracy over time to counteract and tackle threats of a dynamic nature, which is absent in most of the conventional models. However, among the existing ML-based solutions in detecting attack patterns, many of them are computationally expensive, require a long training time, and a considerably large amount of training data-which are seldom available. An aid to this is the association rule learning (ARL) paradigm, whose models are computationally inexpensive and do not require a long training time. Therefore, this paper proposes an ARL-based intelligent Behavioural Trust Model (iBUST) for securing the CPPS. For this intelligent TMM, a variant of Frequency Pattern Growth (FP-Growth), called enhanced FP Growth (EFP-Growth) algorithm is developed by altering the internal data structures for faster execution and by developing a modified exponential decay function (MEDF) to automatically calculate minimum supports for adapting trust evolution characteristics. In addition, a new optimisation model for finding optimum parameter values in the MEDF and an algorithm for transmuting a 1D quantitative feature into a respective categorical feature are developed to facilitate the model. Afterwards, the trust class of an object is identified employing the Naive Bayes classifier. This proposed model is evaluated on a trust evolution-supported experimental environment along with other compared models taking a benchmark dataset into consideration, where it outperforms its counterparts.

The automatic keyphrase extraction techniques aim to extract high-quality keyphrases for document summarization and many other purposes, including indexing and search optimization’ content classification and categorization, topic modeling and analysis, text mining, data analysis, and so on. However, most of the existing techniques are often domain-specific or require domain knowledge. Again, some of these methods employ complex statistical approaches that require significant computational resources. Others depend on large training data sets that may not always be available. To address these challenges, this paper proposes a new keyphrase extraction technique, named ETeKET or Enhanced Tree-Based Keyphrase Extraction Technique, which is domain-independent, requires minimal statistical knowledge, and does not rely on training data. It employs a special variant of binary tree, called Keyphrase Extraction (KePhEx) tree, to extract final keyphrases from candidate keyphrases. Moreover, to assess the degree of cohesiveness of various nodes concerning the root, it also employs a measure, called the Cohesiveness Index (CI), which offers flexibility to the process. Since the generation of final keyphrases depends on the creation of candidate keyphrases, the proposed technique integrates the KeyBERT model to leverage contextual embeddings for effective and efficient candidate keyphrase extraction. This integration enhances adaptability across different text lengths, with dynamic parameter tuning for optimal performance in varied contexts(e.g., short, medium, and long texts). The effectiveness of ETeKET is evaluated using three benchmark datasets, and the experimental results show significant improvement over TeKET, with an F1-score increase of up to 2.61 for SemEval-10, 1.6 for SemEval-2017, and a noteworthy gain of 6.89 for Thesis100 (top 10 results), highlighting ETeKET's enhanced performance, especially in diverse text types.

The Coastal Patrol and Surveillance Application (CPSA) is developed and deployed to detect, track and monitor water vessel traffic using automated devices. The latest advancements of marine technologies, including Automatic Underwater Vehicles, have encouraged the development of this type of applications. To facilitate their operations, installation of a Coastal Patrol and Surveillance Network (CPSN) is mandatory. One of the primary design objectives of this network is to deliver an adequate amount of data within an effective time frame. This is particularly essential for the detection of an intruder’s vessel and its notification through the adverse underwater communication channels. Additionally, intermittent connectivity of the nodes remains another important obstacle to overcome to allow the smooth functioning of CPSA. Taking these objectives and obstacles into account, this work proposes a new protocol by ensembling forward error correction technique (namely Reed-Solomon codes or RS) in Underwater Delay Tolerant Network with probabilistic spraying technique (UDTN-Prob) routing protocol, named Underwater Delay Tolerant Protocol with RS (UDTN-RS). In addition, the existing binary packet spraying technique in UDTN-Prob is enhanced for supporting encoded packet exchange between the contacting nodes. A comprehensive simulation has been performed employing DEsign, Simulate, Emulate and Realize Test-beds (DESERT) underwater simulator along with World Ocean Simulation System (WOSS) package to receive a more realistic account of acoustic propagation for identifying the effectiveness of the proposed protocol. Three scenarios are considered during the simulation campaign, namely varying data transmission rate, varying area size, and a scenario focusing on estimating the overhead ratio. Conversely, for the first two scenarios, three metrics are taken into account: normalised packet delivery ratio, delay, and normalised throughput. The acquired results for these scenarios and metrics are compared to its ancestor, i.e., UDTN-Prob. The results suggest that the proposed UDTN-RS protocol can be considered as a suitable alternative to the existing protocols like UDTN-Prob, Epidemic, and others for sparse networks like CPSN.

Among the tons of articles that are published every year, a considerable number of substandard articles are also published. One of the primary reasons for publishing these substandard articles is due to applying ineffective and/or inefficient reviewer selection processes. To overcome this problem, several reviewer recommender systems are proposed that do not depend on the intelligence of the human selector. However, most of these existing systems do not take the reviewer feedback score or confidence score into consideration during the recommendation process. Therefore, a new reviewer recommender system is proposed in this paper that recommends a set of reviewers to a set of manuscripts with an objective of attaining a high average confidence score taking several constraints into consideration, including a fixed number of reviewers for a manuscript and a fixed number of manuscripts to a reviewer. The proposed system employs a new similarity threshold discovery technique for facilitating the reviewer recommendation process. Again, since there is hardly any dataset exists that satisfies the requirements of the proposed system, a new dataset is prepared by getting the data from various online sources. The proposed system is evaluated by incorporating several existing selection techniques. The experimental results demonstrate that despite employing various selection techniques, the proposed system can assign most of the articles to the prescribed number of reviewers.

Metaheuristic algorithms play a pivotal role in solving complex optimization problems. This study addresses a common challenge faced by such algorithms, i.e., getting trapped in local optima. This problem hampers the efficacy of metaheuristic algorithms in achieving globally optimal solutions. This limitation is also observed in Hybrid Henry Gas Solubility Optimization Algorithm (HHGSO)-a new metaheuristic based algorithm-which employs a cluster-based approach for population management. While this approach fosters exploration by diversifying the search process, it lacks adaptiveness in cluster assignments and the potentiality of hybridization with other algorithms. These shortcomings hinder the algorithm's ability to effectively escape local optima and harness the strengths of diverse optimization techniques. To mitigate these challenges, this study proposed an Enhanced HHGSO (EHHGSO) algorithm, an advanced hybridization of HHGSO. Here, a novel backtracking based technique is incorporated that can detect the trapping into local optima and can escape from it. Empirical findings substantiate the efficacy of EHHGSO in comparison to its ancestor, HHGSO for various benchmark functions.

It is understood that water is the most valuable natural resource and as like wastewater treatment plants are necessary base to control the environmental balance where they are installed. To ensure good quality effluents, the dynamic and complicated wastewater treatment procedure must be handled efficiently. A global interest has been prompted in conservation, reuse, and alternative water sources due to growing treats over water supply scarcity. Water utilities are searching for more efficient ways to maintain their resources globally. The development of machine learning techniques is starting to offer real opportunities to operate water treatment systems in more efficient manners. This paperwork shows research as well as its development work implemented to predict the performance of petrochemical wastewater treatment. The data were used from a reputed chemical plant and the predictive models were developed by implementation of Backpropagation Neural Network using sample datasets with the parameters of wastewater dataset.

With the rapid growth of scientific publications, researchers often find difficulty in discovering appropriate articles that can mitigate the knowledge gaps to understand a target article (a.k.a., base article in this paper). In this case, reference articles can play an important role. It may happen that a researcher may have to read several levels of references, which is challenging since it increases exponentially over levels. This kind of learning method could be considered as the chronological learning. In this paper, a chronological learning supported recommender system is proposed, which utilizes the reference articles of multiple levels for generating a multi-level weighted graph. The weights of the various nodes in this graph are calculated considering three scores, namely lexical similarity, time-aware influence, and node centrality. Among them, the equation for calculating the time-aware influence score is improved by taking citation counts into consideration so that the articles with the higher citation counts receive higher influence scores, which is more practical. From this graph, a chronological path is selected considering a weight-based selection process envisioning mitigating the knowledge gaps to understand a base article. Since the keyphrase extraction plays an important role in this system, various unsupervised keyphrase extraction techniques are evaluated to discover the most suitable relevant technique for the proposed system.

The extraction of high-quality keywords and sum-marising documents at a high level has become more difficult in current research due to technological advancements and the expo-nential expansion of textual data and digital sources. Extracting high-quality keywords and summarising the documents at a high-level need to use features for the keyphrase extraction, becoming more popular. A new unsupervised keyphrase concentrated area (KCA) identification approach is proposed in this study as a feature of keyphrase extraction: corpus, domain and language independent; document length-free; utilized by both supervised and unsupervised techniques. In the proposed system, there are three phases: data pre-processing, data processing, and KCA identification. The system employs various text pre-processing methods before transferring the acquired datasets to the data processing step. The pre-processed data is subsequently used during the data processing step. The statistical approaches, curve plotting, and curve fitting technique are applied in the KCA identification step. The proposed system is then tested and evaluated using benchmark datasets collected from various sources. To demonstrate our proposed approach’s effectiveness, merits, and significance, we compared it with other proposed techniques. The experimental results on eleven (11) datasets show that the proposed approach effectively recognizes the KCA from articles as well as significantly enhances the current keyphrase extraction methods based on various text sizes, languages, and domains.

Due to the exponential growth of information’s and web sources, Automatic keyphrase extraction is still a challenging issue in the current research area. Keyphrases are very helpful for several tasks in natural language processing (NLP) and information retrieval (IR) systems. Feature extractions for those keyphrases execute a vital role in extracting the top-quality keyphrases and summarising the documents at a superior level. This paper proposes a new region-based distance analysis of keyphrases (RDAK) unsupervised technique for feature extraction of keyphrases from articles. The proposed method comprises six phases: data acquisition and preprocessing, data processing, distance calculation, average distance, curve plotting, and curve fitting. At first, the system inputs the collected different datasets to the preprocessing step by employing some text preprocessing techniques. Afterwards, the preprocessed data is applied to the data processing phase, and then after distance calculation, it is passed to the region-based average calculation process, then curve plotting analysis, and afterwards, the curve fitting technique is utilized. Finally, the proposed system has tested and evaluated the performance through implementing them on benchmark datasets. The proposed system will significantly improve the performance of existing keyphrase extraction techniques.

The idea of this study is to validate a list of keywords derived from a scientific article by a domain expert from years of knowledge with prominent document similarity algorithms. For this study, a list of handcrafted keywords generated by Electric Double Layer Capacitor (EDLC) experts are chosen, and relevant documents to EDLC are considered for the comparison. Then, different similarity calculation algorithms were employed in different settings on the documents such as using the whole texts of the documents, selecting the positive sentences of the documents, and generating similarity score with automatically extracted keywords from the documents. The experiment’s outcome provides us with findings that the machine-generated keywords are mostly similar to the curated list by the domain experts. This study also suggests the preferable algorithms for similarity calculation and automated key-phrase extraction for the EDLC domain.

Healthcare 4.0 has revolutionized the delivery of healthcare services during the last years. Facilitated by it, many hospitals have migrated to the paradigm of being smart. Smartization of hospitals has reduced healthcare costs while providing improved and reliable healthcare services. Thanks to the Internet of Healthcare Things (IoHT) based healthcare delivery frameworks, integration of many heterogeneous devices with varying computational capabilities has been possible. However, this introduced a number of security concerns as many secure communication protocols for traditional networks can not be verbatim employed on these frameworks. To ensure security, the threats can largely be tackled by employing a Trust Management Model (TMM) which will critically evaluate the behavior or activity pattern of the nodes and block the untrusted ones. Towards securing these frameworks through an intelligent TMM, this work proposes a machine learning based Behavioral Trust Model (BTM), where an improved Frequent Pattern Growth ${\left(iFP-Growth\right)}$ algorithm is proposed and applied to extract behavioral signatures of various trust classes. Later, these behavioral signatures are utilized in classifying incoming communication requests to either trustworthy and untrustworthy (trust) class using the Naïve Bayes classifier. The proposed model is tested on a benchmark dataset along with other similar existing models, where the proposed BMT outperforms the existing TMMs.

Today globally, coronavirus disease (COVID-19) has infected over more than 81 million people and killed at least 1771K. This is an infectious disease caused by a newly discovered coronavirus. As a result, scientists and researchers around the globe are now trying to find out the path to battle this disease in the most effective way. Chest X-rays are a widely available modality for immediate care in diagnosing COVID-19. Detection and diagnosis of COVID-19 chest X-rays would be more precise for the current situation. In this paper, a phase by phase approach using the concept of one shot learning is introduced for effective classification of chest X-ray images. The proposed method utilizes the application of Entropy for selecting best describing images for effective learning purposes. The proposed model is evaluated on a publically available large dataset of size 24614 images comprising of three classes viz COVID-19, Normal and Non-COVID. The obtained results are promising and encouraging.

Social networking sites have become a daily part of human life. people worldwide use them regularly to interact with each other and stay updated on the most recent news. however, the increasing use of these sites also results in several caveats concerning the security authorities worldwide. one of these concerns is the lack of user awareness, one of the primary reasons behind many recent attacks on social networking sites. while using any of these sites, the users are often required to provide some information, so that website admins can check their user's validity. most of the time, the users have control over this information to decide the audience of their shared data. however, due to a lack of awareness, many users do not look over these privacy settings and end up sharing confidential data publicly. this chapter discusses the significance of user awareness to secure social networking networks within cyberspace.

The recent outbreak of the novel Coronavirus Disease (COVID-19) has given rise to diverse health issues due to its high transmission rate and limited treatment options. Almost the whole world, at some point of time, was placed in lock-down in an attempt to stop the spread of the virus, with resulting psychological and economic sequela. As countries start to ease lock-down measures and reopen industries, ensuring a healthy workplace for employees has become imperative. Thus, this paper presents a mobile app-based intelligent portable healthcare (pHealth) tool, called ${i}$ WorkSafe, to assist industries in detecting possible suspects for COVID-19 infection among their employees who may need primary care. Developed mainly for low-end Android devices, the ${i}$ WorkSafe app hosts a fuzzy neural network model that integrates data of employees’ health status from the industry’s database, proximity and contact tracing data from the mobile devices, and user-reported COVID-19 self-test data. Using the built-in Bluetooth low energy sensing technology and K Nearest Neighbor and K-means techniques, the app is capable of tracking users’ proximity and trace contact with other employees. Additionally, it uses a logistic regression model to calculate the COVID-19 self-test score and a Bayesian Decision Tree model for checking real-time health condition from an intelligent e-health platform for further clinical attention of the employees. Rolled out in an apparel factory on 12 employees as a test case, the pHealth tool generates an alert to maintain social distancing among employees inside the industry. In addition, the app helps employees to estimate risk with possible COVID-19 infection based on the collected data and found that the score is effective in estimating personal health condition of the app user.

Material researchers are progressively embracing the utilization of machine learning techniques to find hidden patterns in data and make predictions without explicit human development. Thousands of papers have been published in the use of carbon for supercapacitor applications. The manufacturing conditions for getting highly super-capacitive carbons from bio-wastes could be analyzed from the existing data using proper machine learning techniques. This work aims to provide a solution called feed forward back propagation neural networks, a supervised learning approach for the prediction of super-capacitive energy storage materials. The proposed method is to apply on the prediction of key parameters with the actual data of the two processes. The configuration of Levenberg-Marquardt backpropagation neural network has been given the smallest mean square error (0.002892, 0.006884) with correlation coefficient (0.992, 0.9789) respectively was three-layer artificial neural network with hidden layer with 9 neurons. The ANN results showed that neural network model can be satisfactorily simulate and predict the behavior of the process.

This paper discusses a new variant of Henry Gas Solubility Optimization (HGSO) Algorithm, called Hybrid HGSO (HHGSO). Unlike its predecessor, HHGSO allows multiple clusters serving different individual meta-heuristic algorithms (i.e., with its own defined parameters and local best) to coexist within the same population. Exploiting the dynamic cluster-to-algorithm mapping via penalized and reward model with adaptive switching factor, HHGSO offers a novel approach for meta-heuristic hybridization consisting of Jaya Algorithm, Sooty Tern Optimization Algorithm, Butterfly Optimization Algorithm, and Owl Search Algorithm, respectively. The acquired results from the selected two case studies (i.e., involving team formation problem and combinatorial test suite generation) indicate that the hybridization has notably improved the performance of HGSO and gives superior performance against other competing meta-heuristic and hyper-heuristic algorithms.

Although, several authentication schemes have already been proposed for smart devices; however, most of these schemes does not consider the fact that smart devices come in different sizes. Hence, they are not screen size independent — which is the point of interest in this paper. Again, alongside screen size independence, a secure scheme also must defend the aforementioned attacks. Taking these concerns into account, in this paper, a hybrid screen size independent authentication scheme is proposed for smart devices that integrates Vibration Code or VC and Press Touch Code or PTC using a juggling-based approach. Here, VC ensures resilience against the shoulder surfing attack since it is a sense based technique and hybridization of both these schemes contribute to attaining resilience against the smudge and brute force attacks up to some extent. In addition, the proposed scheme does not require a space more than the placement of a finger on the screen; and thus, it is screen size independent. The proposed scheme is evaluated in terms of security and functionality; and compared with other similar schemes where it outperforms the others.