Search

The increasing expansion of the industrial and domestic sectors, combined with the integration of renewable energy sources, has led to an overload of the existing interconnected power systems, which in turn has led to significant bottlenecks. This confluence of factors manifests itself in severe frequency, voltage and tie-line power issues, highlighting the urgent need for intelligent control mechanisms. Automatic voltage regulation (AVR) and load frequency control (LFC) are key components that ensure the supply of high-quality energy to consumers while maintaining the nominal frequency, voltage and power deviations in the grid. These measures are essential to ensure the stability and safety of the IPS under these challenging conditions. This research addresses the control strategies for a four-area sophisticated IPS. The complexity of the system, which includes five generation units in each area, including gas, thermal reheating, hydropower and two renewable energy sources (wind and photovoltaic), requires a careful study of the control methods. In particular, the inclusion of various non-linear factors, such as the governor dead band (GDB), generation rate constraint (GRC) and boiler dynamics (BD) in each of the four areas, increases the realism of the study. In this context, a recently introduced meta-heuristic algorithm, the One-to-One-based Optimizer (OOBO), was used to determine the optimal parameters of the proportional-integral-proportional-derivative (PI-PD) controller. The tuning process involves using the integral of time multiplied by the squared error value (ITSE) as the error criterion for evaluating the fitness function. In the study, the voltage, frequency and power performance of the OOBO PI-PD controller is evaluated and compared with alternative control methods, namely Proportional-Integral-Derivative (PID), Integral-Proportional (I-P) and Integral-Proportional-Derivative (I-PD) controllers, all of which are tuned with OOBO. Under the influence of a 5% load step load perturbation (SLP), comprehensive comparisons show the superior performance of the OOBO PI-PD controller, which shows better responses in all areas. Furthermore, the reliability and effectiveness of the OOBO PI-PD method are validated by a sensitivity analysis that considers simultaneous variations of the turbine time constant and speed control within a range of ±25%. The results highlight the robustness of the OOBO-PI-PD control

Biometric authentication technologies, which use physiological and behavioral traits to verify identity, have added a new layer of protection. However, concerns about privacy, security, and illegal entry persist. Therefore, multibiometric systems have emerged to improve authentication accuracy and system robustness. However, to address these problems more effectively, it is imperative to incorporate cancelable biometrics into multimodal systems. This integration is essential because it offers enhanced security measures and safeguards privacy. This paper presents a novel cancelable multibiometric method that utilizes non-negative matrix factorization (NMF) and a lightweight deep learning model to augment the levels of security and privacy in biometric authentication. The proposed system leverages the distinct capabilities of finger vein and fingerprint recognition modalities to enhance authentication accuracy and improve resistance against spoofing attacks. The implementation of cancelable biometrics principles serves to safeguard user privacy and enhance security measures. NMF is employed to facilitate the extraction of features and the obfuscation of data. The study presents actual evidence to support the system’s high validation results with an average accuracy of 95.31% for the NUPT-FPV dataset and 92.71% for the FVC2004 and FV-USM datasets, its ability to preserve anonymity, and its real-time capabilities. Significantly, it satisfies all conditions that can be canceled, representing a noteworthy progression in the domain of cancelable multibiometric. The amalgamation of NMF with a streamlined deep learning model presents a pragmatic and effective resolution, surmounting prior constraints in computational intricacy.

To improve mobility and rehabilitation, precise and adaptive control mechanisms have been developed for lower limb exoskeletons. Brain-computer interface (BCI) provides advanced and intuitive control of assistive and rehabilitation exoskeletons to aid the user. Functional near-infrared spectroscopy (fNIRS) is a non-invasive, and portable brain imaging modality, gained momentum in rehabilitation studies in the last decade. This study provides a novel approach to control a lower limb exoskeleton with enhanced classification accuracy using fNIRS-based BCI, the k-nearest neighbors (kNN) classifier, and optimal feature combination. The brain signals were acquired using fNIRS for walking vs rest for twenty healthy participants, having ten trials for each participant. The statistical measures: mean, peak, variance, skewness, kurtosis, and slope are extracted as features. Optimal feature combination was analyzed and selected for enhanced classification accuracy. kNN was analyzed and selected as an optimal classifier with optimal ‘k’ (number of nearest neighboring data points that the kNN considers while classifying a new data point) using elbow method to improve classification performance. The proposed method achieves an average classification accuracy of $88.19~\pm ~2.55$ %, in offline configuration. In order to control exoskeleton in online settings, simulated online classification was performed using one unknown trial, fed as real-time signal. Sliding window of 2.5 sec is used and achieved average classification accuracy of 97.5%. This research represents a major advancement in user-centric assistive technologies and advances the field of neuro-powered exoskeletons. It also lays the groundwork for future advancements in the integration of neuroimaging, machine learning, and rehabilitation.

In recent years, the proliferation of online communication platforms and social media has given rise to a new wave of challenges, including the rapid spread of malicious bots. These bots, often programmed to impersonate human users, can infiltrate online communities, disseminate misinformation, and engage in various activities detrimental to the integrity of digital discourse. It is becoming more and more difficult to discern a text produced by deep neural networks from that created by humans. Transformer-based Pre-trained Language Models (PLMs) have recently shown excellent results in challenges involving natural language understanding (NLU). The suggested method is to employ an approach to detect bots at the tweet level by utilizing content and fine-tuning PLMs, to reduce the current threat. Building on the recent developments of the BERT (Bidirectional Encoder Representations from Transformers) and GPT-3, the suggested model employs a text embedding approach. This method offers a high-quality representation that can enhance the efficacy of detection. In addition, a Feedforward Neural Network (FNN) was used on top of the PLMs for final classification. The model was experimentally evaluated using the Twitter bot dataset. The strategy was tested using test data that came from the same distribution as their training set. The methodology in this paper involves preprocessing Twitter data, generating contextual embeddings using PLMs, and designing a classification model that learns to differentiate between human users and bots. Experiments were carried out adopting advanced Language Models to construct an encoding of the tweet to create a potential input vector on top of BERT and their variants. By employing Transformer-based models, we achieve significant improvements in bot detection F1-score (93%) compared to traditional methods such as Word2Vec and Global Vectors for Word Representation (Glove). Accuracy improvements ranging from 3% to 24% compared to baselines were achieved. The capability of GPT-4, an advanced Large Language Model (LLM), in interpreting bot-generated content is examined in this research. Additionally, explainable artificial intelligence (XAI) was utilized alongside transformer-based models for detecting bots on social media, enhancing the transparency and reliability of these models.

In the energy sector, the application of renewable energy sources especially solar photovoltaics (PV), is expanding exponentially. Inverters find application in converting DC power from solar PV generating arrays to AC power and feeding it to grid. This study proposes a 5-level switched-capacitor multilevel inverter (SCMLI) that can be used for solar PV applications. The problem of leakage current was mitigated with the use of common ground configuration. Based on various performance parameters a comparison with recently proposed SCMLI’s has been carried out. The proposed SCMLI achieves an efficiency of 95.03% under a 2-kW load. It utilizes one input DC power source, 7 switches, and 2 capacitors only. The experimental results of the proposed inverter show better harmonic profile with reduced Total Harmonic Distortion (THD). The performance of the proposed MLI topology is validated through MATLAB/SIMULINK. Moreover, loss analysis has been carried out on PLECS platform. The hardware implementation is done on a power converter test bed.

Computer-aided decision support systems (DSSs) are becoming popular in a variety of professions. Notably, medical DSSs assist healthcare professionals (decision makers) choose the optimal course of action (decisions) while treating patients. Such systems help decision-makers in situations when there is uncertainty in manual decisions due to lack of information or expertise. Choosing a suitable learning algorithm in a DSS is essential and affects its performance. Among machine learning (ML) algorithms, artificial neural networks (ANNs) are considered the most suitable framework for many classification tasks. In healthcare, an ML-based prediction system/DSS employs data (genetic profile or clinical characteristics) and learning algorithms to forecast target values, which may give promising results. However, improving prediction accuracy is a crucial step in making informed decisions. One can apply various preprocessing methods (cross validation, feature selection, bagging, boosting, etc.) to achieve this. For complex classification tasks like cancer, decision-makers can utilize the hybridization of classifiers to increase prediction accuracy. The presented study investigates the possibilities of improvements in the design of hybridized systems for DSSs to assist healthcare professionals in robust decision-making before, during, and after cancer diagnosis. Since the network weights and the activation functions are the two crucial elements in the learning process of an ANN, this study is organized to investigates the improvement in the hybrid system by selecting suitable features from gene expression microarray data and using these features to compute the more realistic initial weights instead of using random guesses as initial weights for ANN. The use of the proposed framework gives promising results (upto 6.67% gain in accuracy when compared to previous study (Table-5) while 10.43% increase in accuracy when compared to conventional ML classifiers (Table-4).

The adoption of cloud computing has been increasingly common across several industries in recent years and offers unparalleled flexibility and scalability in managing computational resources. However, the increasing reliance on cloud infrastructure also raises concerns regarding the monitoring and mitigation of drift, which pertains to unforeseen modifications inside the cloud environment. The aforementioned modifications possess the capacity to induce operational disruptions and can affect cloud provider reputation. This paper introduces a novel methodology for a proactive drift detection approach, referred to as the Ensemble of Variational Auto-encoder featuring Sequential Contextual Network Extended Kolmogorov-Smirnov Test (EnVACK) a Multi-Mode Variational Auto-encoder with Kolmogorov-Smirnov. It analyzes the data distribution using a non-parametric deep learning approach to create the concept drift map to address sudden and gradual drift challenges in the non-Gaussian cloud domain. This map gives a drift overview at each resource and combined level. This has shown a 96% f-score in the cloud domain an improvement of 4% when evaluated against benchmark studies executed in different domains. The present methodology has been specifically devised to warn and predict the occurrence of drift in synthetic and real-world non-Gaussian cloud usage traces.

Manage Pressure Drilling (MPD) presents a pioneering methodology for enhancing control over well operations and drilling processes, furnishing a proactive approach to mitigate risks stemming from fluctuating downhole pressures. Through the meticulous regulation of drilling fluid via a choke valve, MPD adeptly manages fluid pressure dynamics, effectively addressing challenges posed by undesirable pressure fluctuations. These fluctuations, often instigated by the vertical oscillations of the drill string due to wave disturbances, known as Heave Motion, can jeopardize the integrity of the well. MPD thus serves as a robust contingency against weather-induced delays, optimizing operational efficiency and bolstering the effectiveness of offshore drilling endeavors. This paper introduces various nonlinear control techniques, notably Sliding Mode Control (SMC) and Super-Twisting Sliding Mode Control (STSMC), with a specific emphasis on their application. The central objective is to utilize these controllers to effectively manage the choke valve and mitigate pressure fluctuations within the well. The controller gains underwent detailed refinement using a genetic algorithm, with a focus on minimizing the integral time absolute error (ITAE). The mathematical analysis involves using Lyapunov stability criteria to confirm that the introduced controllers are asymptotically stable. Additionally, comprehensive performance comparisons among the proposed control strategies were meticulously conducted using MATLAB/Simulink simulations. Moving beyond simulation, real-time validation involved an experimental setup integrating the capabilities of the C2000 Delfino with the MCU F28379D Launchpad.

This work addresses the important issue of frequency stability in shipboard microgrids (SMGs), which are confronted with difficulties because of the alternating power insertion from green sources and low system inertia. In order to sustain a consistent frequency (despite system uncertainties), a study and precisely calibrated regulator is necessary. Therefore, this article proposed a new combined three degree of freedom proportional integral derivative (3DOF-PID) and tilted-integral (TI) regulators for achieving enhanced system performance, robustness, and enhance the frequency performance of a power grid. The superior performance of the suggested 3DOF-PID-TI has been assured by comparing it with three distinct controller architectures (FOPID, PDPID2 and 2DOF-PID) for the multi-energy SMG system. The investigation also took into account the temporal delays brought on by the communication links between the sensor and the regulator. The triangulation topology aggregation optimizer (TTAO), a relatively new meta-heuristic technique that hasn’t been employed to load frequency control (LFC) issues until recently, was employed to adjust the controllers’ parameters. The TTAO’s frequency performance outcomes were thoroughly contrasted to those of other optimization algorithms (i.e., Chimp, Whale, and Gradient-Based Optimization Algorithms) so as to evaluate the optimization efficacy of the suggested optimizer. The findings showed that the proposed 3DOF-PID-TI controller, whose parameters has been tuned by TTAO, outperformed its competitors in terms of overshoot (15.87mHz), undershoot (-29.04mHz), settling time (2.49sec) and performance index (i.e., ITAE(0.0171)). Additionally, the suggested controller’s robustness was assessed in a range of SMG operating situations as ex. (Random Multi-Step Variation in Load, Real Data on Stochastic Power Fluctuations, Energy Storage System Impact on the system and sensitivity analysis). The acquired data amply proved that when crucial system parameters experience a substantial variation, the controller’s gains set under normal circumstances do not need to be re-tuned.

Sentiment Analysis, a crucial tool for analyzing user opinions, has shown efficacy particularly when tailored to specific domains. While existing research predominantly focuses on training various classifiers for sentiment analysis within the software engineering (SE) domain, the outcomes often lack consistency when tested across different datasets. To address this gap, this paper proposes a novel approach utilizing transfer learning-based classifiers, fine-tuned and evaluated across diverse SE datasets. A comprehensive study is conducted, benchmarking machine learning and deep learning classifiers for SE sentiment analysis. Results indicate that transfer learning classifiers, namely GPT and BERT, outperform traditional approaches. Notably, the Bert large model achieves an F1-score of 0.89 on the Stack Overflow dataset, surpassing existing state-of-the-art tools. This research not only provides centralized insights but also paves the way for developing more accurate domain-specific sentiment analysis tools tailored for Software Engineering.

In today’s interconnected world, cybersecurity has emerged as a critical domain for ensuring the integrity, confidentiality, and availability of digital assets. Within this sphere, insider threats represent a unique and particularly insidious class of security risks, originating not from external hackers but from within the organization itself. These threats are perpetrated by individuals with inside information concerning the organization’s security practices, data, and computer systems. Traditional security measures like firewalls, intrusion detection systems, and antivirus software are often inadequate for tackling insider threats effectively, owing to their focus on external threats. This inadequacy underscores the urgent need for the development and implementation of more sophisticated, targeted detection techniques for insider threats. In response to this challenge, our research introduces an innovative approach that employs the Density-Based Local Outlier Factor (DBLOF) algorithm, fine-tuned to specifically tackle the challenges posed by the imbalanced nature of the CERT r4.2 insider threat dataset. This dataset is characterized by a highly skewed distribution, with a significant majority of benign instances and only a minimal proportion of malicious activities. Conventional detection algorithms often fail to effectively identify these rare but dangerous instances, leading to a high rate of false negatives. Our methodology capitalizes on the algorithm’s ability to focus on the local density deviation of data points, thereby enabling the precise identification of outliers that are indicative of potential insider threats. Through rigorous testing and validation processes, we have achieved outstanding results, with an of F-score 98%. These remarkable outcomes not only affirm the effectiveness of the DBLOF algorithm as a powerful tool for combating insider threats but also contribute valuable insights to the broader academic and professional discourse on cybersecurity. Importantly, our findings have practical implications, offering organizations actionable recommendations for boosting their internal security mechanisms against the complex and evolving landscape of insider threats.

Microgrids are the future version of advanced distribution networks due to the fast growth of renewable energy resources near consumers’ side. The microgrids are operated in on-grid mode (OGM) with the utility grid, and isolation mode (IM) without the utility grid. This dual operational mode causes protection and control challenges in the microgrids. This research paper suggests an advanced hardware-supported fault detection, phase identification & localization method for AC microgrids. The scheme deploys a Discrete Kalman Filter (DKF) for state estimation of voltage and current signals. Then, a Mathematical Morphology (MM) is engaged for generating a novel fault detection/classification index named segregated energy signature (SES) from estimated voltage and current signals. The system is considered to be faulty if the SES is higher than a predefined threshold setting, while phase identification is achieved by default because of the per-phase implementation of DKF&MM. Moreover, the directional features of the cumulative energy signature (CES) are also computed from MM-based non-fundamental current and voltage to localize the faulty section. The established scheme is tested on the CIGRE microgrid benchmark test bed on Matlab-Simulink software. In addition, the suggested method is also examined on the dSPACE MicroLab testing hardware setup in the Smart grid lab at USPCAS-E, NUST, Pakistan. The result illustrates that the proposed scheme successfully detects, classifies, and localizes the low impedance fault (LIF) as well as high impedance fault (HIF) in both operational modes and topological structures with 96.6% accuracy.

Three frequent factors such as low classification accuracy, computational complexity, and resource consumption have an impact on malware evaluation methods. These challenges are exacerbated by elements such as unbalanced data environments and specific feature generation. To address these challenges, we aim to identify optimal texture features and classifiers for effective malware detection. The article outlines a method that consists of four stages: malware conversion to grayscale, feature extraction using (segmentation-based fractal texture analysis (SFTA), Local Binary Pattern (LBP), Haralick, Gabor, and Tamura), classification using (Gaussian Discriminant Analysis (GDA), k-Nearest Neighbor (KNN), Logistic, Support Vector Machines (SVM), Random Forest (RF), Extreme Learning Machine (Ensemble)), and finally the evaluation. Using the Malimg imbalanced and MaleVis balanced datasets, we assess classifier performance and feature effectiveness. Comparative analysis indicates that KNN outperforms other classifiers in terms of Accuracy, Error, F1, and Precision, while SVM and RF as runners-up. Gabor performs better in MaleVis, whereas the SFTA feature performs better under the Malimg dataset. The proposed SFTA-KNN and Gabor-KNN methods achieve 96.29% and 98.02% accuracy, respectively, surpassing current state-of-the-art approaches. Additionally, higher computing performance is achieved by using fewer dimensions when employing our feature extraction method.

Citrus canker is among the major plant diseases caused by Xanthomonas citri which affects the quality and quantity of citrus fruit. This results in the reduction of citrus production which causes a huge financial loss and livelihood of the farming community. Thus, it is critically important to build a robust, accurate, and time-efficient detection method for real-time identification of the disease. Due to their powerful learning capabilities and improved feature extraction, deep learning approaches have made it feasible to carry out a number of tasks related to the identification of citrus canker in citrus leaves. Previous research has primarily focused on detecting citrus canker on fruits, early detection on leaves can facilitate the adoption of preventive measures before the disease reaches a critical stage. This paper proposes a novel deep learning-based approach for determining the growth rate of citrus canker by classifying it into six distinct stages: water soaking, yellow chlorosis/initiation, chlorosis, blister formation, canker development start, canker infection (50% of the inoculated area), and canker infection (100% of the inoculated area). The proposed approach involves image conversion, size reduction, image augmentation, and the utilization of DenseNet-121. Experimental results demonstrate a classification accuracy of 98.97% using the suggested approach. The Accuracy was 98.97% with macro precision 97%, weighted precision 99%, Macro recall 98%, weighted recall 98%, macro F1_Score 97% and weighted F1_Score 98%. This study presents a unique technique for detecting and classifying the growth rate of citrus canker based on six different stages, while also calculating the temporal change in the affected area of the disease in inoculated citrus leaves. Furthermore, a mathematical model is proposed to predict the disease’s growth rate at any given time, offering valuable insights for disease management and prevention.

Human life has become smarter by utilizing big data, telecommunication technologies, and wearable sensors over pervasive computing to give better healthcare services. Big data is built with the possibility to improve the healthcare industry. Big data makes the interconnection between patients, wearable sensors, healthcare caregivers, and providers through the utilization of Information and Communication Technology (ICT) and software. Most of the economic challenges in developing countries are caused by the healthcare sector, which occurs predominantly due to the increasing population requiring more quality of care concerning older people. Older people need great attention and care as they lead with irreparable damages when a minor accident or insignificant disease occurs. Therefore, the necessity of implementing new technologies and tools has arisen to support senior citizens regarding their healthcare. Various advancements in wireless technology, miniaturization, computing power, and processing made diverse healthcare innovations that led to developing the connected medical devices. Hence, this proposal develops a new healthcare monitoring system for tracking the activities of elderly people, where the Hadoop MapReduce technique for parallel processing the large-sized data. The data collected as mentioned in the available datasets is performed by using the numerous wearable sensors fixed on the “subject’s left ankle, right arm, and chest” that are transformed to the cloud platform and also to the data analytics layer according to the Internet of Medical Things (IoMT) devices. The given input undergoes data splitting to produce tiny chunks. These small chunks of the input files are then considered as Map tasks. Here, in the map phase, the features are optimally selected by the Hybrid Dingo Coyote Optimization (HDCO). The combiner phase classifies the physical activities using the developed Deep Ensemble Learning (DEL) consisting of classifiers such as “Extreme Learning Machine (ELM), deep Convolutional Neural Network (CNN), Long short-term memory (LSTM), Deep Belief Network (DBN), and Deep Neural Network (DNN)”. The parameter tuning in these classifiers is done by the same HDCO. The reducer phase extracts data from different chunks by merging the same classes. The developed HDCO-DEL has secured 13.66%, 16.01%, 17.33%, 13.6%, and 14.01% better accuracy than ELM, CNN, LSTM, DBN, DNN, and HealthFog, respectively on second dataset. The comparison with existing methods shows its better performance and also predicts physical activities with overall high accuracy.

Men’s fertility depends on their bodies making normal sperm and delivering them. Semen analysis has been the test of choice for assessing the male partner in an infertile couple using a single threshold value to distinguish ‘abnormal’ and ‘normal’ parameters. In the semen analysis process, rA©gime issues might affect the semen morphology, quality, and spermatozoa, and also reduce the risks of fertility due to food regimen including glycemic content and limited intake of nutrients. Determination of the connotation between adjustable rA©gime and semen morphology is a complex task to determine fertility. The goal is this study is the prediction of men’s fertility rate to analyze the connection between spermatozoa and the level of lifeblood. Impaired semen parameters alone cannot be used to predict fertility more accurately. Some factors might affect the spermatozoa, like impairing sperm function, morphology, and sustainability, and can reduce the men’s fertility rate. In this article, deep learning on convolutional neural network (DLNN) is used to predict the men’s fertility rate more quickly, accurately, and consistently from different age spam of men between 18–50 years old. The convolutional neural network performs the segmentation of sperm heads, while the deep learning algorithm allows us to calculate the movement speed of sperm heads. After the application of DLNN, we have achieved semen prediction 80.952% and sperm concentration 85.714% accuracy of sperm head detection on human spermatozoa sperm samples. The results of the experiments presented below will show the applicability of the proposed method to be used in automated artificial insemination workflow.

Smart manufacturing is transforming the manufacturing industry by enhancing productivity and quality, driving growth in the global economy. The Internet of Things (IoT) has played a crucial role in realizing Industry 4.0, where machines can communicate and interact in real-time. Despite these advancements, security remains a major challenge in developing and deploying smart manufacturing. As cyber-attacks become more prevalent, researchers are making security a top priority. Although IoT and Industrial IoT (IIoT) are used to establish smart industries, these systems remain vulnerable to various types of attacks. To address these security issues, numerous authentication methods have been proposed. However, many of these methods are vulnerable to known attacks, such as physical security, privileged-insider, and impersonation attacks, or have high computational and communication costs, making them unsuitable for resource-limited IoT devices. Therefore, in this paper, we present a new approach to mutual authentication between the flexible manufacturing system unit, the user, and the server. It enables secure communication in an IIoT-enabled system, which represents the smart manufacturing industry. This security is achieved through the establishment of session keys. Our proposed scheme demonstrates robust resistance against various security attacks, outperforming existing schemes. Although the communication overhead is slightly higher compared to some benchmark schemes, this trade-off is justified by the significant security advantages it offers. Overall, our scheme strikes a balance, providing superior security and competitive performance.

In this paper, a self-excited topology by utilizing third harmonic field excitation is proposed for realizing the brushless operation of the Wound Rotor Vernier Machine (WRVM). The proposed topology consists of two windings housed inside the stator periphery connected in series with the three-phase diode rectifier. The two windings on the stator periphery are the 12-pole excitation and 4-pole armature winding. The 12-pole excitation winding is linked to the open-end star winding of 4-pole armature winding with the help of a three-phase diode rectifier. When the current is supplied by the inverter, a 4-pole, and a 12-pole Magneto Motive Force (MMF) is developed in the machine air gap. The 4-pole fundamental MMF component develops the main stator field and a 12-pole MMF component develops the third harmonic field. The third harmonic field tends to induce a harmonic current in the harmonic winding which is housed inside the rotor periphery. The induced harmonic current is then rectified by the means of an H-bridge rectifier to excite the rotor field winding, and this realizes the brushless operation of a WRVM. The Two-dimensional Finite Element Analysis (2D-FEA) is carried out in a JMAG designer software to validate the operation of the proposed BL-WRVM. Furthermore, the proposed topology is analyzed for the comparative analysis with the sub-harmonic model after the verification of FEA results. The comparative analysis shows that the third harmonic topology results in improved average torque when compared to the subharmonic wound-rotor vernier machine.

The Riva approach is used to develop an object-based Business Process Architecture (BPA) that helps in capturing the full organizational strategies. However, due to the lack of effective tools that can generate Riva BPA models from available artifacts, producing such models is time consuming non-automatic process and it requires an exhaustive manual validation to ensure the development of error-free models. This paper introduces a novel and domain-independent tool titled ‘ARivaT’ to automatically generate Riva BPA models from available knowledge assets such as Units of Work diagrams. ARivaT is underpinned by a step-by-step methodological approach that automates the process of generating Units of Work, First-Cut, and Second-Cut Process Architecture Diagrams. It also provides stakeholders with insightful explanations to allow a precise understanding of business processes and workflow. Furthermore, ARivaT employs a rule-based mechanism to seamlessly validate all the generated Riva BPA models. A case study-based approach has been followed to evaluate the applicability of ARivaT to derive Riva BPA diagrams. The results of our experiment are promising since 82 percent time saving has been recorded when using ARivaT to generate Riva BPA diagrams in comparison to using general drawing tools such as MS Word and MS Visio. In addition, 69 percent time saving has been noted when performing the same task with ARivaT as opposed to specialized tools such as Camunda.

A switched-capacitor based 13-level inverter with high gain and reduced switch stress is presented in this article. The 13-level single-phase ac output voltage is achieved with a voltage gain of 6 by utilizing a single dc source, 14 switches, 3 capacitors and 1 diode in the proposed converter. Self-voltage balancing is achieved for different types of loads without the use of complicated control schemes or supplementary circuitry. The proposed topology can generate negative polarity without a back-end H-bridge. Consequently, voltage stresses across the switches are reduced. The voltage stress across the switches is limited to 4 times the input DC voltage. Included also is an analysis of power loss for several components of the proposed converter. A comparison is made with contemporary topologies that require a single dc source at the input in terms of several performance characteristics. Resistive (R) and resistive plus inductive (R-L) loads are simulated in this study. To demonstrate the effectiveness of the suggested converter, a laboratory prototype is constructed. In order to demonstrate the viability of the suggested arrangement, the experimental results for step variation in load demand and variation in modulation index under practical loading situations are obtained.

This work presents a free-breathing dynamic contrast-enhanced (DCE) MRI reconstruction method called low-rank plus sparse (L+S) with joint sparsity. The proposed method improved dynamic contrast performance by integrating an additional temporal Fast Fourier Transform (FFT) constraint into the standard L+S decomposition method. In the proposed method, both temporal total variation (TV) sparsity constraint and temporal FFT constraint are integrated into a standard L+S decomposition model, forming L+S with joint sparsity. Temporal TV and Temporal FFT aim to suppress under-sampling artifacts and improve dynamic contrast in DCE-MRI, respectively. A fast composite splitting algorithm (FCSA) is adopted for solving the L+S model with multiple sparsity constraints, maintaining the reconstruction efficiency. A computer simulation framework was developed to compare the performance of L+S with joint sparsity and other reconstruction schemes. The performance of L+S with joint sparsity was tested using computer simulation and several liver DCE-MRI datasets. The proposed L+S based method achieved around four times faster reconstruction speed than the GRASP method. With the support of an additional sparsity constraint, the peak DCE signal in the proposed method was increased by more than 20% over that of a standard L+S decomposition.

Design of a vision-based traffic analytic system for urban traffic video scenes has a great potential in context of Intelligent Transportation System (ITS). It offers useful traffic-related insights at much lower costs compared to their conventional sensor based counterparts. However, it remains a challenging problem till today due to the complexity factors such as camera hardware constraints, camera movement, object occlusion, object speed, object resolution, traffic flow density, and lighting conditions etc. ITS has many applications including and not just limited to queue estimation, speed detection and different anomalies detection etc. All of these applications are primarily dependent on sensing vehicle presence to form some basis for analysis. Moving cast shadows of vehicles is one of the major problems that affects the vehicle detection as it can cause detection and tracking inaccuracies. Therefore, it is exceedingly important to distinguish dynamic objects from their moving cast shadows for accurate vehicle detection and recognition. This paper provides an in-depth comparative analysis of different traffic paradigm-focused conventional and state-of-the-art shadow detection and removal algorithms. Till date, there has been only one survey which highlights the shadow removal methodologies particularly for traffic paradigm. In this paper, a total of 70 research papers containing results of urban traffic scenes have been shortlisted from the last three decades to give a comprehensive overview of the work done in this area. The study reveals that the preferable way to make a comparative evaluation is to use the existing Highway I, II, and III datasets which are frequently used for qualitative or quantitative analysis of shadow detection or removal algorithms. Furthermore, the paper not only provides cues to solve moving cast shadow problems, but also suggests that even after the advent of Convolutional Neural Networks (CNN)-based vehicle detection methods, the problems caused by moving cast shadows persists. Therefore, this paper proposes a hybrid approach which uses a combination of conventional and state-of-the-art techniques as a pre-processing step for shadow detection and removal before using CNN for vehicles detection. The results indicate a significant improvement in vehicle detection accuracies after using the proposed approach.

Source code search plays an important role in software development, e.g. for exploratory development or opportunistic reuse of existing code from a code base. Often, exploration of different implementations with the same functionality is needed for tasks like automated software transplantation, software diversification, and software repair. Code clones, which are syntactically or semantically similar code fragments, are perfect candidates for such tasks. Searching for code clones involves a given search query to retrieve the relevant code fragments. We propose a novel approach called Clone-Seeker that focuses on utilizing clone class features in retrieving code clones. For this purpose, we generate metadata for each code clone in the form of a natural language document. The metadata includes a pre-processed list of identifiers from the code clones augmented with a list of keywords indicating the semantics of the code clone. This keyword list can be extracted from a manually annotated general description of the clone class, or automatically generated from the source code of the entire clone class. This approach helps developers to perform code clone search based on a search query written either as source code terms, or as natural language. With various experiments, we show that (1) Clone-Seeker is effective in finding clones from BigCloneBench dataset by applying code queries and natural language queries; 2) Clone-Seeker has a higher recall when searching for semantic code clones (i.e., Type-4) in BigCloneBench than the state-of-the-art; 3) Clone-Seeker is a generalized technique as it is effective in finding clones in Project CodeNet dataset by applying code queries and natural language queries. 4) Clone-Seeker with manual annotation outperforms other variants in finding clones on the basis of natural language queries

Fast access of data from Data Warehouse (DW) is a need for today’s Business Intelligence (BI). In the era of Big Data, the cache is regarded as one of the most effective techniques to improve the performance of accessing data. DW has been widely used by several organizations to manage data and use it for Decision Support System (DSS). Many methods have been used to optimize the performance of fetching data from DW. Query cache method is one of those methods that play an effective role in optimization. The proposed work is based on a cache-based mechanism that helps DW in two aspects: the first one is to reduce the execution time by directly accessing records from cache memory, and the second is to save cache memory space by eliminating non-frequent data. Our target is to fill the cache memory with the most used data. To achieve this goal aging-based Least Frequently Used (LFU) algorithm is used by considering the size and frequency of data simultaneously. The priority and expiry age of the data in the cache memory is managed by dealing with both the size and frequency of data. LFU sets priorities and counts the age of data placed in cache memory. The entry with the lowest age count and priority is eliminated first from the cache block. Ultimately, the proposed cache mechanism efficiently utilized cache memory and fills a large performance gap between the main DW and the business user query.

Deep learning-based plant disease detection has gained significant attention from the scientific community. However, various aspects of real horticultural conditions have not yet been explored. For example, the disease should be considered not only on leaves, but also on other parts of plants, including stems, canes, and fruits. Furthermore, the detection of multiple diseases in a single plant organ at a time has not been performed. Similarly, plant disease has not been identified in various crops in the complex horticultural environment with the same optimized/modified model. To address these research gaps, this research presents a dataset named NZDLPlantDisease-v1, consisting of diseases in five of the most important horticultural crops in New Zealand: kiwifruit, apple, pear, avocado, and grapevine. An optimized version of the best obtained deep learning (DL) model named region-based fully convolutional network (RFCN) has been proposed to detect plant disease using the newly generated dataset. After finding the most suitable DL model, the data augmentation techniques were successively evaluated. Subsequently, the effects of image resizers with interpolators, weight initializers, batch normalization, and DL optimizers were studied. Finally, performance was enhanced by empirical observation of position-sensitive score maps and anchor box specifications. Furthermore, the robustness/practicality of the proposed approach was demonstrated using a stratified k-fold cross-validation technique and testing on an external dataset. The final mean average precision of the RFCN model was found to be 93.80%, which was 19.33% better than the default settings. Therefore, this research could be a benchmark step for any follow-up research on automatic control of disease in several plant species.

Recently many researches have explored the potential of visual programming in robotics, the Internet of Things (IoT), and education. However, there is a lack of studies that analyze the recent evidence-based visual programming approaches that are applied in several domains. This study presents a systematic review to understand, compare, and reflect on recent visual programming approaches using twelve dimensions: visual programming classification, interaction style, target users, domain, platform, empirical evaluation type, test participants' type, number of test participants, test participants' programming skills, evaluation methods, evaluation measures, and accessibility of visual programming tools. The results show that most of the selected articles discussed tools that target IoT and education, while other fields such as data science, robotics are emerging. Further, most tools use abstractions to hide implementation details and use similar interaction styles. The predominant platforms for the tools are web and mobile, while desktop-based tools are on the decline. Only a few tools were evaluated with a formal experiment, whilst the remaining ones were evaluated with evaluation studies or informal feedback. Most tools were evaluated with students with little to no programming skills. There is a lack of emphasis on usability principles in the design stage of the tools. Additionally, only one of the tools was evaluated for expressiveness. Other areas for exploration include supporting end users throughout the life cycle of applications created with the tools, studying the impact of tutorials on improving learnability, and exploring the potential of machine learning to improve debugging solutions developed with visual programming.

A fast method to simultaneously calibrate multiple MEMS Magnetic Inertial Measurement Units (MIMUs) accurately in the field is needed in many application areas. The MEMS MIMUs require calibration of systematic errors of bias, sensitivity, non-orthogonality and misalignment, which vary with temperature and use. Even after calibration, the sensors undergo stochastic errors in static and dynamic conditions and thus uncertainty of output must also be modeled. We propose a method for easy and fast calibration of multiple MIMUs together, while mounted on a single platform. The precise alignment of sensors is not assumed. Our method calibrates both fixed array of MEMS MIMUs or many independent MIMUs simultaneously using kinematic constraints. The novelty of our approach is that the uncertainty of sensors output is also learned as part of our model. Compared with existing state-of-art methods, our algorithm gives more consistent readings of all MIMUs and our framework also predicts the associated uncertainty of the sensor output. The uncertainty prediction of individual sensors is particularly helpful in the sensor fusion.

We propose a deep learning based framework that learns data-driven temporal priors to perform 3D human pose estimation from six body worn Magnetic Inertial Measurement units sensors. Our work estimates 3D human pose with associated uncertainty from sparse body worn sensors. We derive and implement a 3D angle representation that eliminates yaw angle (or magnetometer dependence) and show that 3D human pose is still obtained from this reduced representation, but with enhanced uncertainty. We do not use kinematic acceleration as input and show that it improves the generalization to real sensor data from different subjects as well as accuracy. Our framework is based on Bi-directional recurrent autoencoder. A sliding window is used at inference time, instead of full sequence (offline mode). The major contribution of our research is that 3D human pose is predicted from sparse sensors with a well calibrated uncertainty which is correlated with ambiguity and actual errors. We have demonstrated our results on two real sensor datasets; DIP-IMU and Total capture and have come up with state-of-art accuracy. Our work confirms that the main limitation of sparse sensor based 3D human pose prediction is the lack of temporal priors. Therefore fine-tuning on a small synthetic training set of target domain, improves the accuracy.

Economies of the world and workforce level of skills are changing rapidly. Industrial revaluations and knowledge economies have raised the demand for a knowledgeable and digitally equipped skilled workforce. Technical and Vocational Education and Training (TVET) is an educational stream that provides vocational skills to youth and produces a skilled workforce for the industry. The role of Information and Communication Technologies (ICT) in the TVET pedagogy, training delivery, teachers training, curriculum, and labs is need of the day to meet current era challenges. In the TVET education training cycle, the level of ICT enablement is required to be reviewed. This Systematic Literature Review (SLR) evaluates literature for innovation of ICT in TVET education for the last ten years. This study aims to identify and present ICT-based technology innovations, research, and applications used in TVET training cycle system components/functional areas to identify gaps for future research directions/agenda. This gap identification will help (i) TVET training institutes to upgrade their teachers, curriculum, labs and, equipment, (ii) policymakers to devise technology-oriented skilled workforce development policies, (iii) to provide guidelines to industry, researchers, and academics to focus on gaps to address future challenges of workforce development. Based on the search string, we found 2,445 relevant documents and after applying quality assurance and inclusion/exclusion criteria, finally, 134 documents were selected for the study and analyzed. The result of this systematic literature review identifies that ICT technologies and application dispersion into TVET training cycle system components/functional areas is very low particularly in monitoring and evaluation, career guidance and job placement, trainee’s assessment, and teacher’s training. TVET Technology index suggests that much focus is needed on IoT, Robotics, Data Science, Artificial intelligence, cloud computing, and other similar technology induction to all TVET training.

Smart cities and their applications have become attractive research fields birthing numerous technologies. Fifth generation (5G) networks are important components of smart cities, where intelligent access control is deployed for identity authentication, online banking, and cyber security. To assure secure transactions and to protect user’s identities against cybersecurity threats, strong authentication techniques should be used. The prevalence of biometrics, such as fingerprints, in authentication and identification makes the need to safeguard them important across different areas of smart applications. Our study presents a system to detect alterations to biometric modalities to discriminate pristine, adulterated, and fake biometrics in 5G-based smart cities. Specifically, we use deep learning models based on convolutional neural networks (CNN) and a hybrid model that combines CNN with convolutional long-short term memory (ConvLSTM) to compute a three-tier probability that a biometric has been tempered. Simulation-based experiments indicate that the alteration detection accuracy matches those recorded in advanced methods with superior performance in terms of detecting central rotation alteration to fingerprints. This makes the proposed system a veritable solution for different biometric authentication applications in secure smart cities.

Pulsed electric field (PEF) pasteurisation keeps treated liquid food fresh and nutritious compared to traditional thermal pasteurisation. However, PEF adoption is still limited on an industrial scale due to a lack of practical systems. As a result, a great deal of research has gone into overcoming the limitations of the existing systems. Keeping this in mind, the current study contributes to the improvement of the electroporator. The heterogeneous electric field’s distribution raises the temperature of the treated food samples. Liquid laminar flow is a reason for heterogeneous electric field’s distribution in continuous treatment. Hotspots may also be created by using an inefficient high-voltage waveform in addition to the heterogeneous electric field distribution. This study rectifies the heterogeneous distribution by proposing an improved coaxial treatment chamber and double-exponential waveform to replace the exponential-decaying waveform. A static mixer provides an increased mixing, i.e. disrupting the laminar flow, inside the treatment zone. COMSOL based computational model was developed to study flow behaviour and corresponding temperature distribution in the proposed coaxial treatment chamber with sieves. Based on the model, it has been concluded that coaxial electrodes with sieves provide more homogeneous flow properties inside the treatment chamber. The effectiveness of the double-exponential (DE) waveform was validated using MATLAB. A three-stage Marx generator giving the DE waveform was designed and constructed. The performance of the improved treatment chamber together with the DE waveform, known as the electroporator, was studied using chemical and microbial analysis. Untreated, PEF treated, and thermal treated orange samples were stored at 4°C for 9 days before being examined. The lowest microbial growth was observed in both the PEF treated with sieves and thermally treated food samples than the untreated sample. However, treated juices’ visual and chemical colour analysis showed that the PEF-treated sample acquired a brighter appearance than a thermally processed sample. Thus, this study provides significant findings in developing and utilising an electroporator to inactivate microorganisms.

Due to the sheer global energy crisis, concerns about fuel exhaustion, electricity shortages, and global warming are becoming increasingly severe. Solar and wind energy, which are clean and renewable, provide solutions to these problems through distributed generators. Microgrids, as an essential interface to connect the power produced by renewable energy resources-based distributed generators to the power system, have become a research hotspot. Modern research in the field of microgrids has focused on the integration of microgrid technology at the load level. Due to the complexity of protection and control of multiple interconnected distributed generators, the traditional power grids are now outmoded. Microgrids are feasible alternatives to the conventional grid since they provide an integrating platform for micro-resources-based distributed generators, storage equipment, loads, and voltage source converters at the user end, all within a compact footprint. A microgrid can be architected to function either in grid-connected or standalone mode, depending upon the generation, integration potential to the main grid, and consumers’ requirements. The amalgamation of distributed energy resources-based microgrids to the conventional power system is giving rise to a new power framework. Nevertheless, the grids’ control, protection, operational stability, and reliability are major concerns. There has yet to be an effective real-time implementation and commercialization of micro-grids. This review article summarizes various concerns associated with microgrids’ technical and economic aspects and challenges, power flow controllers, microgrids’ role in smart grid development, main flaws, and future perspectives.

This article proposes boosting the multiplication performance for convolutional neural network (CNN) inference using a precision prediction preprocessor which controls various precision approximate multipliers. Previously, utilizing approximate multipliers for CNN inference was proposed to enhance the power, speed, and area at a cost of a tolerable drop in the accuracy. Low precision approximate multipliers can achieve massive performance gains; however, utilizing them is not feasible due to the large accuracy loss they cause. To maximize the multiplication performance gains while minimizing the accuracy loss, this article proposes using a tiny two-class precision controller to utilize low and high precision approximate multipliers hybridly. The performance benefits for the proposed concept are presented for multi-core multi-precision architectures and single-core reconfigurable architectures. Additionally, a design for a merged reconfigurable approximate multiplier with two precisions is proposed for utilization in single-core architectures. For performance comparison, several segments-based approximate multipliers with different precisions were synthesized using CMOS 15nm technology. For accuracy evaluation, the concept was simulated on VGG19, Xception, and DenseNet201 using the ImageNetV2 dataset. This article will demonstrate that the proposed concept can achieve significant performance gains with a minimal accuracy loss when compared to designs that utilize exact multipliers or single-precision approximate multipliers.

The application of cloud computing has diversified with the adoption of Internet of Things (IoTs) and edge computing. However, it has increased the uncertainty of workload demand; thus, the efficient utilization of cloud computing resources become more challenging. Traditionally, dynamic consolidation of workload inside cloud data centers relies on identifying overload and under-load hosts using either static or dynamic threshold value. In this paper, we propose a Utilization Driven Model (UDM) model to estimate the number of under-utilized and over-utilized processing machines through percentile ranks of low and high utilization from mean value of resource utilization of hosts and the value of mean absolute deviation of resource demand. UDM swiftly reacts to any change in workload demand and adapts the system to the current demand of resource utilization. The UDM approach not only impacts the energy consumption and quality of service but also increases the elastic nature of cloud by robustly managing the sudden changes in workload. Experiment results show that UDM is an efficient server consolidation technique, improving 30% energy, 40% quality of service compared to contemporary techniques. Thus, the UDM is more robust to support stochastic resource demand compared to traditional techniques.

A two-way tunable power divider for Simultaneous Wireless Information and Power Transfer (SWIPT) is proposed in this paper. The power divider divides the incoming power in such a way that maximum power is utilized for energy harvesting purposes and the rest of the power is used for information decoding purposes. This unequal distribution of power is realized by tuning the varactor diode's reverse biasing voltages. A good matching performance is attained using the radial and short stubs at the input and one of the output ports, respectively. High isolation is achieved with the help of the resistor used for connecting the output ports of the power divider. The achieved power division ratios vary from 13:1 to 28:1. The tunable power divider is designed and fabricated at 2.4 GHz. The bandwidth measured at 10 dB return loss is 3.2 GHz and the isolation better than 12 dB is achieved between the output ports. The designed structure is compact and good consistency is observed between the simulated and measured results. The designed power divider may be used for future 5G wireless networks.

In this paper, we introduce a nonlinear vibrations analysis and dynamic responses of a vertical conveyor system under multi excitation forces. By adding the nonlinear proportional derivative controller (NPD) to the vibrating motion of the vertical vibration conveyor, the energy was transferred between uncontrolled and controlled system. We calculate the approximate solutions of the vibrating system utilizing the method of multiple scales. In addition, we investigate the stability at worst resonance cases using phase plane technique, equations of frequency response, and averaging method. The vertical vibration conveyor behavior was studied numerically at the values of its different parameters. The results exhibit the efficiency of NPD control unit to avoid the oscillations of the vertical conveyor system. Numerical simulations have been carried out using MAPLE and MATLAB software's to ensure the fidelity of our results. Comparisons are made between analytical and numerical results. Also, findings of the present work are discussed in details and compared with published works.

Pattern recognition (PR) algorithms have shown promising results for upper limb myoelectric control (MEC). Several studies have explored the efficacy of different pre and post processing techniques in implementing PR-based MECs. This paper explores the effect of segmentation type (disjoint and overlap) and segment size on the performance of PR-based MEC, for multiple datasets recorded with different recording devices. Two PR-based methods; linear discriminant analysis (LDA) and support vector machine (SVM) are used to classify hand gestures. Optimum values of segment size, step size and segmentation type were considered as performance measure for a robust MEC. Statistical analysis showed that optimum values of segment size for disjoint segmentation are between 250ms and 300ms for both LDA and SVM. For overlap segmentation, best results have been observed in the range of 250ms-300ms for LDA and 275ms-300ms for SVM. For both classifiers the step size of 20% achieved highest mean classification accuracy (MCA) on all datasets for overlap segmentation. Overall, there is no significant difference in MCA of disjoint and overlap segmentation for LDA (P-value = 0.15) but differ significantly in the case of SVM (P-value

This paper surveys the optimization frameworks and performance analysis methods for large intelligent surfaces (LIS), which have been emerging as strong candidates to support the sixth-generation wireless physical platforms (6G). Due to their ability to adjust the behavior of interacting electromagnetic (EM) waves through intelligent manipulations of the reflections phase shifts, LIS have shown promising merits at improving the spectral efficiency of wireless networks. In this context, researchers have been recently exploring LIS technology in depth as a means to achieve programmable, virtualized, and distributed wireless network infrastructures. From a system level perspective, LIS have also been proven to be a low-cost, green, sustainable, and energy-efficient solution for 6G systems. This paper provides a unique blend that surveys the principles of operation of LIS, together with their optimization and performance analysis frameworks. The paper first introduces the LIS technology and its physical working principle. Then, it presents various optimization frameworks that aim to optimize specific objectives, namely, maximizing energy efficiency, sum-rate, secrecy-rate, and coverage. The paper afterwards discusses various relevant performance analysis works including capacity analysis, the impact of hardware impairments on capacity, uplink/downlink data rate analysis, and outage probability. The paper further presents the impact of adopting the LIS technology for positioning applications. Finally, we identify numerous exciting open challenges for LIS-aided 6G wireless networks, including resource allocation problems, hybrid radio frequency/visible light communication (RF-VLC) systems, health considerations, and localization.

The combination of renewable energy sources and prosumer-based smart grid is a sustainable solution to cater to the problem of energy demand management. A pressing need is to develop an efficient Energy Management Model (EMM) that integrates renewable energy sources with smart grids. However, the variable scenarios and constraints make this a complex problem. Machine Learning (ML) methods can often model complex and non-linear data better than the statistical models. Therefore, developing an ML algorithm for the EMM is a suitable option as it reduces the complexity of the EMM by developing a single trained model to predict the performance parameters of EMM for multiple scenarios. However, understanding latent correlations and developing trust in highly complex ML models for designing EMM within the stochastic prosumer-based smart grid is still a challenging task. Therefore, this paper integrates ML and Gaussian Process Regression (GPR) in the EMM. At the first stage, an optimization model for Prosumer Energy Surplus (PES), Prosumer Energy Cost (PEC), and Grid Revenue (GR) is formulated to calculate base performance parameters (PES, PEC, and GR) for the training of the ML-based GPR model. In the second stage, stochasticity of renewable energy sources, load, and energy price, same as provided by the Genetic Algorithm (GA) based optimization model for PES, PEC, and GR, and base performance parameters act as input covariates to produce a GPR model that predicts PES, PEC, and GR. Seasonal variations of PES, PEC, and GR are incorporated to remove hitches from seasonal dynamics of prosumers energy generation and prosumers energy consumption. The proposed adaptive Service Level Agreement (SLA) between energy prosumers and the grid benefits both these entities. The results of the proposed model are rigorously compared with conventional optimization (GA and PSO) based EMM to prove the validity of the proposed model.

There has been a rapid increase in dietary ailments during the last few decades, caused by unhealthy food routine. Mobile-based dietary assessment systems that can record real-time images of the meal and analyze it for nutritional content can be very handy and improve the dietary habits and, therefore, result in a healthy life. This paper proposes a novel system to automatically estimate food attributes such as ingredients and nutritional value by classifying the input image of food. Our method employs different deep learning models for accurate food identification. In addition to image analysis, attributes and ingredients are estimated by extracting semantically related words from a huge corpus of text, collected over the Internet. We performed experiments with a dataset comprising 100 classes, averaging 1000 images for each class to acquire top 1 classification rate of up to 85%. An extension of a benchmark dataset Food-101 is also created to include sub-continental foods. Results show that our proposed system is equally efficient on the basic Food-101 dataset and its extension for sub-continental foods. The proposed system is implemented as a mobile app that has its application in the healthcare sector.

A multimodal biometric system integrates information from more than one biometric modality to improve the performance of each individual biometric system and make the system robust to spoof attacks. In this paper, we propose a secure multimodal biometric system that uses convolution neural network (CNN) and Q-Gaussian multi support vector machine (QG-MSVM) based on a different level fusion. We developed two authentication systems with two different level fusion algorithms: a feature level fusion and a decision level fusion. The feature extraction for individual modalities is performed using CNN. In this step, we selected two layers from CNN that achieved the highest accuracy, in which each layer is regarded as separated feature descriptors. After that, we combined them using the proposed internal fusion to generate the biometric templates. In the next step, we applied one of the cancelable biometric techniques to protect these templates and increase the security of the proposed system. In the authentication stage, we applied QG-MSVM as a classifier for authentication to improve the performance. Our systems were tested on several publicly available databases for ECG and fingerprint. The experimental results show that the proposed multimodal systems are efficient, robust, and reliable than existing multimodal authentication systems.

Putting trust in the world of the Internet of Things, where served and serving entities are often unknown, is very hard especially when personal and business information is often being exchanged for providing and consuming services. Moreover, the issues of interoperability and scalability of billions of heterogeneous things in the IoT systems require more attention. A user-centric model of complex interconnected things must be designed in a way that not only makes things trustworthy for common people but it also provides the solution for interoperability and scalability. ARCA-IoT is such a system which not only identifies the attributes (including quality of service) essential for trust but also presents a user-centric model that is robust enough to tackle the attacks made by dishonest entities to manipulate the trustworthiness. For scalability and interoperability, a cloud-assisted environment is introduced in the ARCA-IoT. An intuitive Naive Bayes approach is used to train the ARCA-IoT in a way that it calculates the probabilities of the trustworthiness of the entities and then identifies various types of attacks with the support of three proposed algorithms. The approach is validated with a specifically designed simulated environment. Based on our simulation results, the ARCA-IoT demonstrates the effectiveness in term of performance metrics such as accuracy, sensitivity, specificity, and precision. Besides this, the system outperforms the existing related approaches in terms of a qualitative analysis based on different parametric metrics such as interoperability, scalability, context-awareness, and a human-like decision.

Cancer is the second leading cause of mortality across the globe. Approximately 9.6 million people are estimated to have died due to cancer disease in 2019. Accurate and early prediction of cancer can assist healthcare professionals to devise timely therapeutic innervations to control sufferings and the risk of mortality. Generally, a machine learning (ML) based predictive system in healthcare uses data (genetic profile or clinical parameters) and learning algorithms to predict target values for cancer detection. However, optimization of predictive accuracy is an important endeavor for accurate decision making. Reject Option (RO) classifiers have been used to improve the predictive accuracy of classifiers for cancer like complex problems. In a gene profile all of the features are not important and should be shaved off. ML offers different techniques with their own methodology for feature selection (FS) and the classification results are dependent on the datasets each having its own distribution and features. Therefore, both FS methods and ML algorithms with RO need to be considered for robust classification. The main objective of this study is to optimize three parameters (learning algorithm, FS method and rejection rate) for robust cancer prediction rather than considering two traditional parameters (learning algorithm and rejection rate). The analysis of different FS methods (including t-test, Las Vegas Filter (LVF), Relief, and Information Gain (IG)) and RO classifiers on different rejection thresholds is performed to investigate the robust predictability of cancer. The three cancer datasets (Colon cancer, Leukemia and Breast cancer) were reduced using different FS methods and each of them were used to analyze the predictability of cancer using different RO classifiers. The results reveal that for each dataset predictive accuracies of RO classifiers were different for different FS methods. The findings based on proposed scheme indicate that, the ML algorithms along with their dependence on suitable FS methods need to be taken into consideration for accurate prediction.

The patient of Parkinson's disease (PD) is facing a critical neurological disorder issue. Efficient and early prediction of people having PD is a key issue to improve patient's quality of life. The diagnosis of PD specifically in its initial stages is extremely complex and time-consuming. Thus, the accurate and efficient diagnosis of PD has been a significant challenge for medical experts and practitioners. In order to tackle this issue and to accurately diagnosis the patient of PD, we proposed a machine-learning-based prediction system. In the development of the proposed system, the support vector machine (SVM) was used as a predictive model for the prediction of PD. The L1-norm SVM of features selection was used for appropriate and highly related features selection for accurate target classification of PD and healthy people. The L1-norm SVM produced a new subset of features from the PD dataset based on a feature weight value. For the validation of the proposed system, the K-fold cross-validation method was used. In addition, the metrics of performance measures, such as accuracy, sensitivity, specificity, precision, F1 score, and execution time, were computed for model performance evaluation. The PD dataset was in this paper. The optimal accuracy achieved the best subset of the selected features that might be due to various contributions of the PD features. The experimental findings of this paper suggest that the proposed method can be used to accurately predict the PD and can be easily incorporated in healthcare for diagnosis purpose. Currently, the computer-based assisted predictive system is playing an important role to assist in PD recognition. In addition, the proposed approach fills in a gap on feature selection and classification using voice recordings data by properly matching the experimental design.