Computing, Internet of Things and Data Analytics

The incorporation of AI techniques has assumed a critical role in enhancing marketing strategies and ensuring sustained business growth. AI, armed with well-curated data and effective training, exhibits the capability to anticipate consumer preferences and deliver recommendations, thereby enabling precise marketing. Through the application of AI methods, precision marketing enables personalized interactions between businesses and customers. It attracts potential clients, offers tailored marketing recommendations to high-value customers, and ultimately reduces marketing costs. Thus, this paper aims to conduct a comprehensive analysis of the chosen methods using a predefined set of criteria. The comparative study will commence by introducing each method, followed by the selection of criteria for comparison. To obtain results, a multi-criteria comparative methodology has been adopted, which aligns perfectly with our objectives. The outcomes obtained from this evaluation will expose the strengths and weaknesses of each algorithm, thereby identifying areas for potential future improvements.

The success of the IoT applications in the field of Information Technologies has led to its spread to different areas of use. In this study, the performance management of infrastructures developed using Industrial IoT sensors is examined. First, the impact of performance management in different sectors is explained. Then, different definitions and types of maintenance in the literature are explained comparatively. Then, big datasets obtained with Industrial IoT sensors and their applications are discussed. Also, studies conducted in different fields with different methods using trend monitoring applications are mentioned. The common aspect of these studies is that they provide applications that increase performance management through trend monitoring. The combination of all these concepts and technologies represents a positive effect. For the sustainability of this effect and performance management, decision making strategies in the predictive maintenance of the devices are mentioned.

With the developing technology, the e-commerce market is growing day by day. As of 2022, it is estimated that 19.7% of the sales in the world are made over the internet. However, there are negative elements that distinguish online sales from regular sales. Communication between the seller and the customer is more difficult on the online platform. Likewise, problems such as quality or cargo are constantly written under the product reviews. For this reason, the seller must constantly monitor customer feedback and take the necessary action. With topic modeling algorithms, user complaints can be grouped and read in groups. In this study, LDA (Latent Dirichlet allocation), NMF (Non-Negative Matrix Factorization) and BERTopic algorithms tested on Amazon product review dataset were compared. According to the results obtained, all 3 algorithms are successful and useful. The BERTopic algorithm produced more meaningful results than other algorithms according to the consistency calculation metric.

In the new digital age, the determination of emerging topics has become a central issue for academia. Latent Dirichlet Allocation (LDA) method, a key mechanism for determining trends, has long been a method of great interest in a wide range of fields. However, the probabilistic structure leads to a serious effect on the score not only by changing the parameters but also from trial to trial for the fixed parameters. This study describes the implementation of the LDA method for exploring trend topics in the hydrology domain. Several parameters like the portion of the corpus, the number of topics as well as hyperparameters of the LDA method, $$\alpha $$ α and $$\beta $$ β , have been considered. The emerging topics of the field have been obtained using the parameters which attained the highest mean coherence score.

To solve the issue of battery charge-discharge and associated damage brought on by incorrect estimates of the battery efficiency, fuzzy logic is used to define a new quantity known as the Energy storage system (ESS), which is based on the battery state, state of charge (SOC), and state of health (SoH). A battery management system (BMS) technique is necessary for energy storage systems (ESSs) for ageing increases a battery’s internal resistance and reduces its capacity. To control the battery state using fuzzy logic, in this paper, a formula for calculating battery efficiency is proposed. The charging time, charging current, and battery capacity are all factors in the proposed fuzzy logic-based battery efficiency estimation formula. The findings show that the ESS is used by the fuzzy logic battery management system to determine battery efficiency. The battery efficiency is also decreased by using a defect diagnosis algorithm to construct a safe system when charging and discharging. Applying the proposed BMS algorithm in a 3-kW ESS shows that it is valid.

This study provides an overview of the potential uses of unmanned aerial vehicles and their swarm systems in future military technology. By examining the usage purposes in different application areas, the basic requirements of a common UAV design are investigated. Critical requirements have been determined, and it has been demonstrated that UAV systems with certain key features can be easily incorporated into swarm systems and used in future military applications. In addition, a common UAV that can be utilized in various applications has been developed, and constraints in the design and production stages have been determined. In this respect, the study provides a basis for similar studies to be carried out.

Colorectal cancer is the third most common cancer diagnosed worldwide. The early detection of this disease can help in treating cases and save human lives. Deep learning algorithms appear as interesting tools used to successfully detect and segment polyps, thus improving surgical resection. When implementing these algorithms, there are different architectures to use. Depending on the architecture used, the obtained accuracy and the convergence speed of the algorithms can be different. To assess how these algorithms model datasets, we can use different metrics. In this paper, we compare different segmentation loss function metrics using one conventional architecture called U-Net. The performance was evaluated on three well-known polyp datasets, namely CVC-ClinicDB, Kvasir, and ETIS-Larib PolypDB. Findings show that the best results from CVC-ClinicDB are 85.89% for Dice, 84.89% from Kvasir, and 77.02% from ETIS-Larib PolypDB. The model behaves well even if we combine the three datasets. In fact, the accuracy level still reached 76.71%.

In this work, we will propose a new color image encryption technique using chaotic maps and based on an improved Feistel scheme. This scheme is reinforced by two genetic operators adapted to cryptography. After putting the original image into a single vector, a genetic mutation is applied. Then, a subdivision into sub-blocks of pseudo-random size is performed. Feistel's technique is applied using new confusion and diffusion functions. A genetic crossover is installed at the output and is controlled by a pseudo-random crossover table. Simulations carried out on a large sample of images of various sizes demonstrate that our method is resilient against all known attacks.

This paper presents an algorithm for determining the maximum area or perimeter of a simple or convex polygon with k sides located within a region of interest. Inscribed polygons are used to detect, classify, and segment objects in images. Traditional algorithms generate basic shapes such as triangles, quadrilaterals, and pentagons. However, a user may be interested in obtaining polygons with the maximum area or perimeter, deciding whether they must be simple or convex, and even specifying the type of polygon needed (triangle, quadrilateral, etc.). The problem of determining the maximum area or perimeter of a simple k-sided polygon within a region of interest has not been solved in a generic way for a user-defined value of k. This paper presents a flexible algorithm that allows the user to determine which polygons should be calculated. The algorithm’s source code in C++, Java and Python is available in a GitHub repository to ensure its usability for the scientific community.

Forecasting electricity consumption is a key mechanism for a wide range of industries to develop strategies or take precautions. This case study primarily aims to predict the electricity consumption of the machines on the production line through sensors and analyzers containing both thermocouples and devices storing the use of electricity. To do so, one of the most powerful methods, which is known for its ability to learn the main characteristics of the data for time-series models, the Prophet method has been utilized. Moreover, the capability of the Prophet method relying on both the use and not the use of temperature data in forecasting electricity consumption has been discussed. The achievement of the method has been supported through the tables and figures in both univariate and multivariate cases. Comparing the RMSE, MAE, and SMAPE scores, the results have shown that the not use of temperature data has been better than those of the use of temperature in the prediction.

India has witnessed a notable upsurge in floods owing to shifts in global climatic patterns underlining climate change and global warming. This inorganic change is significantly evident in places like Kerala and Tamil Nadu. Kerala receives perennial rainfall throughout the year, influenced by southeast and northwest monsoon rainfall cycles, resulting in flood-induced calamities like landslides prevailing in the Thrissur and Munnar regions of Kerala. Long-Term Short Memory (LSTM) networks outperform pre-existing deep learning forecasting models as it is optimal for time series forecasting by handling non-linear spatiotemporal dynamics, and adapting to long-term dependencies in time series data while ensuring high scalability. The objective of the present work is rainfall prediction in flood-prone regions in Kerala. We propose an LSTM network to predict the monthly rainfall in Thrissur, Pathanamthitta, Munnar and Kottayam. The proposed model is evaluated using the metrics like mean absolute error and root-mean-squared error.

Males and females inherently perceive cognitive load differently due to their biological make-up. Research indicates that males outperform females in spatial tasks. There is an increased demand for solutions to minimise the gap in visuospatial abilities between sexes. Augmented reality (AR) techniques offer a range of options to minimise the inter-sex gap in visuospatial perception and cognitive processes. However, research on the causes of cognitive differences between sexes in visuospatial tasks is obscure. Studies have shown that males have better reaction time and accuracy than females when performing spatial visualisation and orientation tasks with AR. There are no investigations on the factor of sex-specific differences that might impact user performance. Hence, this research focuses on inter-sex differences in perceptions of AR solutions and their effect on user reaction time and accuracy. The study employs an alternative research design with spatial AR – involving object-identification processing and spatial processing – to research the design criteria for AR solutions. The goal is to reduce participants’ cognitive load and reaction time and, ultimately, increase their performance.

Brain diseases, which encompass a wide range of conditions and illnesses brought on by stroke, Alzheimer’s, Parkinson’s, traumatic brain injury, and many other conditions, afflict 1 in 6 individuals worldwide. One such disorder that gradually destroys nerve cells in the midbrain is Parkinson’s disease. It is a neurodegenerative condition that impairs motor abilities. Additionally, it has an impact on the muscles involved in speech production, leading to hypokinetic dysarthria, a set of motor speech disorders that includes dysphonia, bradylalia, and poor articulation accuracy. In this context, this work proposes the application of recurrence plots to map the speech signals onto images, which will be further fed into a Convolutional Neural Network for the automatic classification of PD from healthy controls. The proposed approach is assessed on the Italian Voice and Speech Data, containing Diadochokinetic (DDK) recordings of /pa/ & /ta/ audio recordings. The experimental results of the full audio file approach produced an average testing accuracy of 83%. Also, the model based on the frame-based approach performed well on the test set, resulting in an average test accuracy of 91% for both /pa/ and /ta/ recordings.

For the foreseeable future, managing the work from home (WFH) workplace will continue to be a challenge for enterprises. An essential stage in developing and improving a plan is utilizing topic modeling and polarity sentiment analysis. This strategy can assist executives in sustaining productivity, customizing employee settings and experiences, and sustaining successful teamwork. The objective of this research is to utilize Reddit’s subreddits and behavioral analytics to detect work-from-home trends in the context of COVID-19, employing advanced computational methods. On textual data collected from 7 communities focusing on WFH-related discussion on Reddit from December 1, 2020, to August 31, 2022, we use sentiment analysis and latent topic modeling. Polarity analysis revealed that communities had a more favorable attitude towards WFH than a negative one. Community members were found to be worried about a range of issues associated to remote employment, according to topic modeling.

As data generated by the IoMT devices are offloaded to the cloud, processing and analyzing such large-scale data to extract useful information presents challenges in terms of latency, bandwidth, and privacy. Edge computing has emerged as a promising paradigm to address these challenges. For this purpose, the proposed approach explores the benefits of performing Feature Selection methods, including filters, wrappers, and embedded techniques, along with ML algorithms at the edge. As a result, a novel framework is proposed that requires an optimal number of features from the dataset to build an optimal data classification model. The results of the study show that feature selection can significantly improve the classification accuracy and performance of ML algorithms when applied to the medical dataset. Moreover, the simulation results confirm that the XGBoost classifier utilizing the ET algorithm achieved the highest classification accuracy of 95% surpassing the current state-of-the-art.

This paper presents a classification model that predicts delays in Indian lower courts based on case information available at filing. The model is built on a dataset of 4.2 million court cases filed in 2010 and their outcomes over a 10-year period. The data set is drawn from 7000+ lower courts in India. The authors employed AutoML to develop a multi-class classification model over all periods of pendency and then used binary decision forest classifiers to improve predictive accuracy for the classification of delays. The best model achieved an accuracy of 81.4%, and the precision, recall, and F1 were found to be 0.81. The study demonstrates the feasibility of AI models for predicting delays in Indian courts, based on relevant data points such as jurisdiction, court, judge, subject, and the parties involved. The paper also discusses the results in light of relevant literature and suggests areas for improvement and future research. The authors have made the dataset and Python code files used for the analysis available for further research in the crucial and contemporary field of Indian judicial reform.

In this article, we will explore a new technique for encrypting color images. This approach is based on the classical Vigenère technique which has been improved and adapted to image encryption. This approach involves the use of two substitution tables of the same size, generated from the chaotic maps. Then, a crossover operation is applied on the integrity of the vector leaving the substitution phase, this operation is controlled by a pseudo-random vector.The standard security tests applied to our technique proves the robustness and effectiveness of our approach against known attacks.

Steganography is the name given to the science of hiding information that has been studied throughout history. With the digitalization of information, traditional physical steganography techniques have been replaced by digital steganography. After the success of the Bitcoin payment system, the potential of blockchain technology, which is the creative technology behind Bitcoin, has been better understood, and it has started to be tried in the field of steganography. In the study, a secured blockchain image-steganography method, which is difficult to detect with steganalysis methods, is proposed. Solana blockchain platform was chosen for the demonstration of the proposed method. In the study, a stego-image was produced from a selected cover-image to hide data by applying the LSB steganography technique, and this stego-image was minted as a Stego-NFT using the Solana platform. The data to be hidden is encrypted with the OTP encryption algorithm. After the transfer of the minted Stego-NFT to the recipient’s wallet, the confidential data transmission process is completed with the extraction of the hidden data and then the burning of the transferred NFT. Thanks to the NFT-based image-steganography (NBS) method proposed in the study, the stego-image is only exposed to a stego-only attack during the period between minting and burning. In other words, since only the stego-object is accessible for steganalysis and the stego-NFT is unique, it cannot be detected by steganalysis techniques such as PSNR and histogram analysis. Finally, with these features, the proposed NBS method is one of the first academic studies in this field in the literature.

Air pollution poses a significant environmental challenge, adversely affecting the health of millions worldwide. Consequently, accurate prediction of pollutant levels has become increasingly crucial to prevent and mitigate the negative impacts of air pollution. This research introduces a Python-based artificial network neural algorithm for predicting PM2.5 levels in Medellín, Colombia, leveraging meteorological and emission data. The model utilizes a Multilayer Perceptron Neural Network, incorporating principal component analysis (PCA) and K-means clustering to determine the optimal number of hidden layers and neurons. Additionally, trend and correlation analyses were conducted to identify the most relevant predictors by examining the relationship between available variables and the target variable (PM2.5). Model performance is assessed using Mean Square Error and Mean Absolute Error.

This paper deals with the estimation of unconditional and conditional Granger-causality spectrum in the frequency domain. We describe two Python routines that parallel the existing R routines in computing these two quantities via package grangers. We present a simulation study showing that under zero-causality processes Python routines tend to perform slightly better than R, while under low-causality processes R routines perform quite better, because Python is less sensitive than R to small causality parameters. This difference can be attributed to the intrinsic VAR order selection procedure of the two packages.

This study attempts to secure the stored images and restrict unauthorized individuals from accessing them to effectively protect the data of the various systems, whether it be an image, a video, or a text file. For the data encryption process, symmetric encryption algorithms have been proposed. The findings of two well-known encryption algorithms were compared to make sure the best methods were being employed. Given that both Blowfish and AES symmetric encryption use block ciphers with a lot of data, they were chosen. Both techniques are capable of encoding high-resolution facial images, and the encoded files can be analyzed using quantitative metrics like histograms and time elapsed with volume scaling. According to the results obtained using the suggested criteria, AES is preferred since it produces specific results in terms of picture and file encoding quality and accuracy, processing and execution speed, coding complexity, coding efficiency, and homogeneity. In conclusion, symmetric encryption methods protect the face recognition system faster and more effectively than alternative options. The AES algorithm is favored over others because of its enormous block space, encryption accuracy, and speed of implementation.

As most disasters are weather-related, location tracking is crucial during any event, and early warning systems can save lives. This article outlines the design and execution of a wireless emergency alerts model for catastrophic alerting based on data from distributed databases by using machine learning enabled Application Programming Interface (API) services. Using machine learning-based historical and current data from the NASA Open API and OpenWeatherMap API, this program is intended to forecast natural catastrophes and aid in their management. To forecast future disasters, the application analyses data from previous natural disasters, including floods, wildfires, tornadoes, cyclones, and hurricanes. Additionally, it gives users access to GPS (Global Positioning System) based evacuation routes and emergency shelter information. The purpose of this article is to assist individuals in the lead-up, during, and aftermath of a disaster by providing access to important resources and information. The software serves as a daily catastrophe management tool, notifying users of the most recent disasters and letting them know whether an incident is happening right now using machine learning-based data. This application can be an important tool to help individuals from disaster.

The storage of oil products is characterized by a high level of risk that can be illustrated by an explosion, fire, or spills from the storage area. As a result, monitoring those areas becomes an essential task for stakeholders especially in countries with a wide storage area. This task cannot be monitored manually as this methodology requires a high number of employers and workers. This work proposes a smart IoT-based approach for real-time monitoring and controlling oil storage areas with an evaluation of the risk level. The proposed approach is then modeled and simulated using multi-agent system (ABM) to represent the dynamic behavioral of the oil storage system in normal and degraded mode. Experiments done have demonstrated the effectiveness of ABM in dynamic modeling, the reliability of the proposed system in monitoring and controlling oil storage facilities and emphasizing the advantages of incorporating IoT in oil storage management.

A fault is a typical state that alters the functioning of the power grid. An abrupt alteration in the current level is noted in the event of a fault due to the incorporation of the Electric Vehicle Charging Station. The distribution system is subject to various types of faults, including open circuit faults and short circuit faults such as L-L-L-G, L-G, L-L-L, and L-L-G faults. The majority of traditional over-current relays are capable of detecting these categories of faults. The over-current relays exhibited a deficiency in detecting high-impedance faults, as their fault current was found to be lower than the standard current value. The presence of high impudence faults can result in the occurrence of arcs and the potential for fire arcing within the distribution system. This results in significant loss of both property and human life. Accurately and promptly detecting these types of faults is imperative. This paper employs a new approach utilizing RADWT & SVM to identify and categories high-impedance faults in the electrical distribution network. The utilization of the rational dilation wavelet transform is employed for the purpose of feature extraction, while the support vector machine is utilized for the detection and classification of faults.

Voice disorders are a widespread issue affecting people of all ages, and accurate diagnosis is crucial for effective treatment. With the recent development of artificial intelligence-based audio and speech processing, research on detection and classification of voice disorders has increased. However, existing work has mostly focused on the binary (two class) classification of voice disorders. Some researchers have also explored multi-class classification, but their results are not promising. In this paper, a framework is proposed for the multi-class classification of voice disorders using OpenL3 embeddings. A pre-trained OpenL3 model is utilized to extract high-level embedding features from the mel spectrogram. Then different classifiers are evaluated after the neighbourhood component analysis (NCA) based feature selection. Random Forest (RF), Support Vector Machine (SVM) and K-Nearest Neighbors (KNN) are employed separately to classify the selected features. The evaluation and comparison are performed on a balanced subset of the Saarbruecken voice database (SVD). Without any speech enhancement preprocessing, our best model, OpenL3-KNN improves the existing work accuracy by 4.9% and F1 score by 8.7%.

Coronary heart disease leading up to stenosis, the partial or total blocking of coronary arteries, is the leading cause of death worldwide. Multi-vessel coronary artery disease affecting two or more coronary requires interpretive expertise on the assessment, the process of interpreting is complex and a time-consuming. Auto-mated identification and classification of angiograms with blockage detection from minimally invasive procedures would be of great clinical value. This study aims at OpenCV method with the help of adaptive thresholding, and brightness corrections for the segmentation of the blood vessels and used the same masked images for the detection of blockage in x-ray angiograms using deep neural nets. The proposed model have achieved 97% accuracy on the task of classifying the x-ray angiogram to ‘Blockage Detected’ and ‘Normal’, with a F1-Score of 0.9532. These results open the way to a fully automated method for the identification of Blockage from X-ray angiograms.

Advancements in Big Data, Robotics, Artificial Intelligence (AI), and the Internet of Things (IoT) have led to the emergence of humanoid robots, set to become commonplace in various aspects of our lives. Through machine learning algorithms and AI, machines can now learn from experiences and respond contextually. This has created a demand for real-time decision-making and assistance systems in healthcare. The paper proposes a new system composed of three components: an assisting subsystem, a decision-making subsystem, and the actors (patients and doctors). One integral part of this architecture is the HIYAM robot, a humanoid robot under development in Morocco. HIYAM is designed to learn, assist patients in medical facilities and homes, and make decisions based on different situations. The robot possesses diverse features such as conversing with patients, walking, searching for medications and definitions, entertaining patients, providing care, and detecting faces and emotions. Developing HIYAM is a complex endeavor that involves enhancing both the assisting and decision-making subsystems. Overall, this paper highlights the progress in humanoid robot technology, particularly the HIYAM robot, and its potential to revolutionize healthcare by assisting patients and supporting decision-making processes.

The forecast combination for energy prices has generally focused on the combination of neural network algorithms. This paper extends the literature by combining conventional models with neural network models. This work analyzes the point forecast combination applied to the daily prices of natural gas in Spain during 2023. To do this, a model with a deterministic trend and ARIMA process in the residuals (RMSE = 1.40) is combined with a neural networks model (RMSE = 1.65) offering excellent results in terms of predictive ability (RMSE = 0.88) in the short term. The best forecasting performance was obtained by applying the Bates and Granger combination method, estimating very similar weights to the two estimated models: 0.58 and 0.42, respectively. The presented combined model outperforms the models reported in the literature. The result of the forecast exercise indicates that the fall in the price of natural gas will continue.

Classification methods use image object features to distinguish between objects and assign them to classes. In the present study we develop a convolutional neural network (CNN) optimized to classify images with embedded vector fields (VFs), generated on the solution $$\hat{u}(x,y)$$ u ^ ( x , y ) of the Poisson equation, which contains the image function in its right-hand side. The embedded VF features subject to extraction, by our CNN, are trajectories and singular points (SP), which augment the image object features. The aim of this paper is to validate that the set of augmented image features increases the separability of the image objects and improves the classification statistics. To reach the aim, we implement our CNN along with four contemporary CNNs to classify two public image databases COIL100 and ISIC2020 as well as their derivatives with embedded VFs. The obtained results are presented in the paper and confirm that embedding VFs with real and complex SPs increases the classification statistics.

Undoubtedly breast cancer identifies itself as one of the most widespread and terrifying cancers across the globe. Millions of women are getting affected each year from it. Breast cancer remains the major one for being the reason of largest number of demise of women. In the recent time of research, Medical Image Computing and Processing has been playing a significant role for detecting and classifying breast cancers from ultrasound images and mammograms, along with the celestial touch of deep neural networks. In this research, we focused mostly on our rigorous implementations and iterative result analysis of different cutting-edge modified versions of EfficientNet architectures namely EfficientNet-V1 (b0-b7) and EfficientNet-V2 (b0-b3) with ultrasound image, named as CEIMVEN. We utilized transfer learning approach here for using the pre-trained models of EfficientNet versions. We activated the hyper-parameter tuning procedures, added fully connected layers, discarded the unprecedented outliers and recorded the accuracy results from our custom modified EfficientNet architectures. Our deep learning model training approach was related to both identifying the cancer affected areas with region of interest (ROI) techniques and multiple classifications (benign, malignant and normal). The approximate testing accuracies we got from the modified versions of EfficientNet-V1 (b0- 99.15%, b1- 98.58%, b2- 98.43%, b3- 98.01%, b4- 98.86%, b5- 97.72%, b6- 97.72%, b7- 98.72%) and EfficientNet-V2 (b0- 99.29%, b1- 99.01%, b2- 98.72%, b3- 99.43%) are showing very bright future and strong potentials of deep learning approach for the successful detection and classification of breast cancers from the ultrasound images at a very early stage.

Collaborative Filtering with clustering is the primary method of recommendation. This research work introduced a new Spectral Bi-clustering with the Mahalabonis distance-based Movie Recommendation algorithm for Collaborative Filtering. In this paper, we compare the performance of several clustering methods like Kmeans, Spectral clustering with radial Basis Function and nearest neighbors affinity, Spectral clustering with radial Basis Function and nearest neighbors affinity with Mahalabonis distance, Spectral Bi-clustering with Mahalabonis distance to generate a movie recommendation system. Our experimental results show a significant performance improvement of our proposed Spectral Bi-clustering with the Mahalanobis distance-based Movie Recommendation algorithm over the traditional K-means algorithm as well as the Spectral Clustering algorithm with nearest neighbors and radial Basis Function affinity for Movie Recommender System.

The study aimed to design and implement a system that provides visitors of Pacaya Samiria Amazon Lodge Private Reserve in the Amazon Jungle of Peru with real-time location-based information and interactive experiences using their smartphones. The proposed system consists of Bluetooth Low Energy (BLE) beacons placed along the Ecotrails that communicate with visitors’ smartphones, providing them with information through AR about the flora, fauna, and cultural heritage. The study utilized a qualitative method approach that involved a focus group. The results suggest that using BLE beacons can significantly enhance visitors’ experiences in ecotourism by providing contextual and immersive information and creating a sense of engagement with the natural and cultural surroundings. The study also highlights the importance of designing ecologically sustainable technology solutions, supporting the long-term preservation of the fragile ecosystem of the Amazon Jungle.

In recent years, business intelligence has been a mechanism that provides important decision-making information to ensure firms’ sustainability and add value to stakeholders. This study provides a review of literature that investigates how and why firms can use resource-based approach to identify factors for business intelligence success. A qualitative approach was adopted to carry out a content analysis of existing literature. The findings indicated that factors like firm’s tangible and intangible resources are associated with information technology and organizations’ system, which supports enhancing firm’s business intelligence capabilities and performance. Based on the literature review findings, a conceptual model has been developed to be empirically tested in future studies. The consolidated findings may contribute to existing literature on business intelligence and resource-based approach may be used in identifying a firm’s success factors in the context of business intelligence.

In this study, we aimed to classify breast cancer patients into four molecular subtypes: Luminal A, Luminal B, Her-2, and triple negative, using six machine learning techniques: logistic regression (LR), naive Bayes (NB), k-nearest neighbors (KNN), support vector machine (SVM), decision tree (DT), and random forest (RF). We evaluated the performance of each model using several evaluation metrics, including accuracy, precision, recall, F1-score, and area under the receiver operating characteristic curve (AUC). The dataset used in this study was obtained in real-time from breast cancer patients, and includes immunohistochemistry (IHC) marker reports. Our results show that all six models achieved high accuracy and AUC scores, indicating their effectiveness in classifying breast cancer patients into molecular subtypes. However, the random forest model outperformed the other models with an AUC score of + 0.95, followed by Logistic Regression with an AUC score of 0.91. These findings demonstrate the potential of machine learning techniques in accurately classifying breast cancer patients into molecular subtypes, which could inform clinical decision-making and personalized treatment strategies.

A systematic study for human-computer interface (HCI) for emotion recognition is presented in this paper, with a focus on various methods used to identify and interpret human emotions. It delves into various methods used to identify and interpret human emotions and highlights the limitations of unimodal HCI for emotion recognition systems. The paper emphasizes the benefits of multimodal HCI and how combining different types of data can lead to more accurate results. Additionally, it highlights the importance of using multiple modalities for emotion recognition. The study has significant implications for mental health assessments and interventions as it offers insights into the latest techniques and advancements in emotion recognition. Future research can use these insights to improve the accuracy of emotion recognition systems, ultimately leading to better mental health assessments and interventions. Overall, the paper provides a valuable contribution to the field of HCI and emotion recognition, and it underscores the importance of taking a multimodal approach for this critical area of research.

Explainable artificial intelligence (XAI) helps users understand the logic behind machine learning model (ML) predictions so that they can better understand and believe model predictions. Many studies have looked at the interaction between humans and XAI, focusing mainly on metrics such as interpretability, fidelity, transparency, trust and usability of explanations. This paper aims to conduct a user study to explore how different types of explanations in the field of XAI affect people’s understanding and behaviour in decision-making. In behavioural science, nudges and boosts are competing approaches and allow a choice architecture to improve decision-making. In our study, we utilized two types of explanations in XAI as a choice architecture, and unveiled the impactful effects of these explanations on behaviour, resulting in alternative decision-making outcomes. Explanations containing actionable information were found to be more effective and understandable. However, our findings indicate that the information provided by certain XAI techniques may not sufficiently persuade users to understand and trust the explanations offered.

The random forest algorithm is a popular machine learning technique that is widely used for classification and regression tasks. Although it is known for its high accuracy and robustness, one of the main challenges associated with the random forest algorithm is its long execution time, particularly when dealing with large datasets. Therefore, several methods have been existed to reduce the execution time of the random forest algorithm, including optimization of hyperparameters, feature selection method and parallelization techniques. In order to do that, the Random Forest (RF) parameters have been proposed, and the experiments for tuning four RF parameters are illustrated based on CPU/GPU with PC and Nvidia Jetson board. In this paper, the results of execution time with Windows and Linux operating systems are presented. NVIDIA’s Jetson platform offers great potential for embedded machine learning, aiming to strike a harmonious balance between the objectives of high accuracy, throughput and high performance. We discussed the weakness and strength for each params, and provided insights into their implementation and performance. The results of Jetson Nano board shown that the proposed methods can significantly reducing the execution time of RF params. All the coding steps are available at https://github.com/HayderMosaMerza/image_registration .

IoT has gained considerable momentum over the years. The development of standards and protocols has enabled the growth of huge and varied applications. Starting with an overview of the current landscape of IoT communication protocols, the study highlights the significance of protocols as a means of enabling seamless connectivity and data exchange between IoT devices. It then examines the opportunities and challenges of IoT-based 6G, featuring existing IoT communication and messaging protocols. Further, the paper dives into technologies and architectures that support IoT communication and messaging, while surveying Artificial Intelligence paradigms backing prospective applications. This investigation supplies a foundation for ongoing research and development designed not only to enhance IoT communication, but also to drive coverage and maximize the potential of IoT enabled applications across a range of industries.

Blind Community faces significant challenges in independently navigating their surroundings, often relying on white canes or assistance from others. These traditional methods can limit their mobility and autonomy. Therefore, there is a pressing need for innovative solutions that leverage emerging technologies to enhance the navigation capabilities of blind individuals. Deep learning techniques have shown great promise in various computer vision tasks, making them a compelling approach to addressing the complexities of blind navigation. In this study, we are proposing a unified approach that provides blind assistance using obstacle avoidance, terrain awareness and person identification. We utilised U-FCHarDNet model and achieved the state-of-the-art mIOU of 67.82. With a processing time of only 20 ms, our approach is well-suited for real-time environments. By leveraging these advancements, we aim to provide efficient navigation assistance and support to the blind community.

Securing Medicare data sharing is of paramount importance. Our current centralized approach to data management falls short of preserving privacy and ensuring safe data sharing. Blockchain technology is explored as a viable solution designed to reinforce security in shared data. We conducted a focused survey to uncover the deficiencies and challenges in existing protocols, which provide valuable insights for the future application of blockchain in patient data exchange. The healthcare sector necessitates immediate data access and entry, yet the risk of data breaches looms large. Our study emphasizes blockchain’s potential to enhance data security, granting patients greater control over their shared data. This study is a stepping stone towards a new epoch of data sharing strategies centered on privacy.

In this paper, we introduce an innovative technique to enhance the performance of Generative Adversarial Networks (GANs) for image generation tasks by incorporating Morphological Neural Networks (MNNs). The primary objective of this research is to address some of the common limitations associated with GANs, such as mode collapse and training instability, by harnessing the distinctive capabilities of MNNs in analyzing and processing image structures. We put forth a methodology that integrates morphological operations, including dilation and erosion, within the generator and discriminator components of GANs. Our experiments are carried out using the CIFAR-10 data-set, and the performance of our proposed integrated model is compared with multiple established GAN variants. The experimental outcomes reveal that our approach significantly improves image quality, convergence, and stability while maintaining a high level of resistance to noise and artifacts. This study lays the groundwork for further investigation into the synergy of MNNs and GANs across a broad spectrum of image generation applications, presenting valuable avenues for future research.

In today's date, fog computing is on the rise due to its characteristics. Its ability to provide smart devices with close-by computing capabilities offers a drastic reduction in cloud traffic and more efficient data transfer. However, data security in fog computing devices is a major concern. Homomorphic Encryption (HME) Scheme is a technique that protects private data from various threats, and to improve efficiency and reduce time, it offers modification of original data after encryption without decrypting it. In this paper, we explore the various homomorphic encryption schemes that can be implemented in fog computing to protect the data and provide data security. We present a detailed architecture and various characteristics of fog computing to better visualize the use of homomorphic encryption in different areas. Further, we present the applications of homomorphic encryption in different sectors and discuss future research directions.

Healthcare is a rapidly expanding field with a substantial amount of heterogeneous data driving innumerable health-related tasks. Various healthcare service providers still rely on manual procedures, which can be time-consuming and require significant effort. To automate such manual operations, recent technological advances have emerged in the domain of Machine Learning (ML). ML falls under the subject of Artificial Intelligence (AI), and it gets combined with ‘big data’ to draw meaningful insights. The integration of ML in the healthcare sector has optimized decision-making and predictive analysis. This paper discusses the various application areas of ML in healthcare. Additionally, several ML algorithms used by other researchers in healthcare-related experiments are summarized. A brief review is provided regarding the experiments. This paper delineates the challenges associated with using ML in healthcare. Finally, the paper offers insights into future research directions.

Cardiac imaging is crucial in the diagnosis of cardiovascular disease. Cardiovascular disease is the umbrella term for the majority of heart ailments. The majority of the causes of mortality are associated with cardiovascular illness. The authors provide a technique for the diagnosis of cardiac disease. The main aim of this study is to determine the most effective technique for predicting cardiovascular disease, specifically focusing on the use of signs of heart disease and Electrocardiogram images. This will be achieved by leveraging the latest advancements in Deep Learning and Machine Learning methods. The authors conduct a comprehensive examination of various Machine Learning and Deep Learning Techniques. These techniques were evaluated in the context of predicting cardiovascular disease evaluating Image. The analysis shows that the Convolutional Neural Network methods are much more effective than the alternatives.

In recent years, the popularity of Short Message Service (SMS) on mobile phones has surged due to technological advancements and the growing prevalence of content-based advertising. However, this has also led to a surge in spam SMS, which can inundate one’s handset at any time and potentially result in the theft of personal information. Researchers have explored a range of options to combat spam SMS, including content-based machine learning algorithms and stylometric approaches. While filtering spam emails has proven to be effective, detecting spam SMS presents a unique set of challenges due to the presence of idioms, abbreviations, and well-known terms and phrases that are frequently used in legitimate messages. This study aims to examine and assess different classification techniques by utilizing datasets gathered from previous research studies. The primary emphasis is on comparing conventional Machine Learning (ML) methods. This study specifically investigates the efficacy of classification algorithms, including Logistic Regression (LR), Naïve Bayes (NB), and Random Forest (RF), in accurately identifying spam SMS messages. The results of our research demonstrate that the RF classifier outperforms other traditional ML techniques in the detection of spam SMS messages. This paper thoroughly examines diverse classification techniques employed in spam detection. The knowledge derived from this study has the potential to advance the development of more effective systems for detecting spam SMS messages. Ultimately, this would enhance the security and privacy of mobile phone users.

Computer vision systems have made advancements in object detection using Artificial Intelligence. This paper presents a comprehensive approach utilizing traditional machine learning models such as decision trees, support vector machines, logistic regression, k-nearest neighbors, and naive Bayes. These models are trained on a publicly available dataset from Github, offering diverse objects for recognition. Practical guidelines are provided for easy experimentation. Evaluation metrics include accuracy, precision, recall, and the F1 score. This paper serves as a valuable resource for object identification in the field.