2690-6791ISSN (Online)
2769-786XISSN (Print)
As energy efficiency and sustainability become paramount in the face of growing urbanization and environmental concerns, predictive energy management in smart buildings has emerged as a promising avenue for mitigating energy consumption and optimizing resource utilization. In this paper, we investigate the application of advanced machine learning techniques, particularly a multi-layer Long Short-Term Memory (LSTM) model, within the framework of the Internet of Things (IoT), to predict and manage energy consumption. We rigorously evaluate our approach against a suite of machine learning baselines, including Linear Regression, Random Forest, Support Vector Machine, and Gradient Boosting, utilizing a comprehensive dataset encompassing power consumption data from smart home appliances and associated weather variables. Our experimental results demonstrate the superior predictive capabilities of the LSTM model, showcasing its ability to outperform traditional machine learning baselines across various metrics, including Mean Absolute Error (MAE), and Root Mean Squared Error (RMSE). These findings underscore the potential of deep learning models in capturing intricate temporal dependencies within energy consumption data, contributing to improved energy efficiency, cost savings, and environmental sustainability in smart building environments. The integration of predictive energy management models into IoT-enabled smart buildings holds the promise of a more intelligent and sustainable future in urban development and resource management.
Read MoreDoi: https://doi.org/10.54216/JISIoT.100201
Vol. 10 Issue. 2 PP. 08-17, (2023)
In the rapidly evolving landscape of the Visual Internet of Things (VIoT), this paper presents a pioneering approach to distributed facial expression recognition—an intelligent system that holds transformative potential for security, human-computer interaction, and personalized services. Our journey unfolds with the development of the Light Vision Transformer (LVT) model, specifically engineered to operate on the resource-constrained edges of the VIoT network. Differentially private federated training ensures both the model's prowess and the preservation of user privacy. Through meticulous experimental evaluations, we validate the effectiveness and efficiency of our approach, shedding light on its scalability and ethical implications. This work is more than a technical endeavor; it symbolizes a commitment to responsible AI, balancing innovation with the preservation of individual rights. Our findings resonate beyond facial expression recognition, serving as a beacon for the VIoT community to explore the dynamic interplay between distributed computing, edge intelligence, and ethical considerations. As we stride towards a more connected and responsive world, this research paves the way for continued exploration, propelling VIoT technology towards a future that is both intelligent and ethically attuned.
Read MoreDoi: https://doi.org/10.54216/JISIoT.100202
Vol. 10 Issue. 2 PP. 18-23, (2023)
Data analysis is an essential component of decision support in various industries that includes industrial and educational institutions. This research proposes Data Mining (DM) techniques to improve the efficiency of higher education (HE) institutions. DM has a substantial impact on different higher education activities including student performances, management of student’s life cycle, selection of courses, monitoring of retention rate, grants & funds management by using technique’s such as clustering, decision trees (DT), and association. Educational Data Mining (EDM) is an interdisciplinary study topic that focuses on getting DM to the fields of education by leveraging methods from (ML) statistics, (DM), and (DA) to get important insights from educational sets of data. EDM is critical in transforming raw data into useful information, allowing for a greater knowledge of students and their academic settings, as well as promoting better teacher assistance and ESD (Educational System Decisions). The study's goal is to provide a complete overview of EDM (Educational Data Mining), highlighting its various applications and benefits in the context of higher education.
Read MoreDoi: https://doi.org/10.54216/JISIoT.100203
Vol. 10 Issue. 2 PP. 24-37, (2023)
Chronic kidney disease (CKD) is a common and possibly fatal condition affecting billions worldwide. Early detection and accurate diagnosis of CKD are critical for timely intervention and improved patient outcomes. In recent years, machine learning techniques have shown great promise in assisting medical professionals in detecting and diagnosing various diseases. This study aims to develop a novel machine learning (ML) model for detecting CKD using clinical and demographic data. The dataset used in this study comprises a comprehensive collection of patient records, including laboratory test results, medical history, and demographic information. Feature selection is one of the techniques that, combined with the ML approach, select the significant features. Several ML algorithms were implemented to detect CKD in the early stages but identified the issues with existing ML algorithms. The developed models' performance is assessed using precision, accuracy, and recall metrics. Additionally, feature importance analysis is conducted to identify the key factors influencing CKD diagnosis. The strength of the proposed approach shows accurately by identifying the individuals at risk of CKD and distinguishing between different stages of the disease. The dataset used for this research was collected from the UCI repository, which consists of 25 attributes, 550 samples, 400 CKD affected, and 150 standard models. The dataset consists of two folders, training and testing. The training utilizes 1000 samples with detailed patient health conditions. The developed CKD detection model shows promising results, achieving high accuracy of 97.98%. on the test dataset. By leveraging machine learning algorithms, this approach can assist healthcare professionals in making more informed decisions regarding early intervention and personalized treatment plans for patients with CKD. Ultimately, applying machine learning techniques in CKD detection can improve patient outcomes and reduce healthcare costs.
Read MoreDoi: https://doi.org/10.54216/JISIoT.100204
Vol. 10 Issue. 2 PP. 38-48, (2023)
Urban environments with high industrialization are infested with hazardous chemicals and airborne pollutants. These pollutants CO, O3, SO2, NO2, and PM can have devastating effects on human health, causing both acute and chronic diseases such as respiratory infections, lung cancer, and heart disease. Air pollution monitoring is vital to warn citizens of the health risks associated with exposure to high concentrations of these criteria pollutants. This study designed a low-cost IoT monitor to measure concentration levels of criteria pollutants emitted from transportation sources within Kwame Nkrumah University of Science and Technology environs. Three monitoring sites, KNUST Tech junction, Ayeduase gate junction and KNUST campus junction, were identified as the locations within the proximity of the university for the deployment of the monitor. Hourly and mean daily CO, NO2, O3 and SO2 concentrations at each of the three sites were measured for a week using the IoT monitor, when students were in school and when students were on vacation. The average daily CO, NO2 and O3 concentrations measured at the selected locations when school was in session and during vacation were presented on histogram. The mean weekly concentrations of CO, NO2 and O3 were also estimated as 13.2ppm, 0.277ppm and 0.106ppb respectively at KNUST Tech junction; 10.1ppm, 0.254ppm and 0.110ppb respectively at Ayeduase gate junction; and 8.0ppm, 0.415ppm and 0.100ppb respectively at the KNUST campus junction when school was in session. The results show that the concentrations of all the pollutants were higher and exceeded the EPA standards except for CO at KNUST Campus junction monitoring site. These high levels of emissions are an indication of a health concern for the students at the university and university authorities can device means of curbing it.
Read MoreDoi: https://doi.org/10.54216/JISIoT.100205
Vol. 10 Issue. 2 PP. 49-62, (2023)
In recent years, most of the research exhibits in the field of Education 4.0 Training Systems (ETS) and Industry 4.0 Training Systems (ITS) that has the ability to learn the behavior of the learners, interns, or trainees. Understanding the feelings and emotions of learners toward learning is essential for creating a practical and exciting learning experience. Patience-emotions detection and sentiments analysis have emerged as an integral part of the understanding of the behaviors of learners, thus there is a need to expand the overall educational or training process in academics and industries. This model enables teachers, trainers and instruction designers to obtain valuable information, which can be used to optimize teaching strategies to improve learning outcomes. To achieve this goal with IoT-enabled objects, an Academician can create a more personalized and effective learning environment for students, trainees and interns. A novel emulated framework is designed and implemented with IoT and machine learning techniques to analyze the performance of learners. The model receives feedback from 1000 learners using IoT devices and analysis the missing information in the learning systems, that missing information lacked effective learning. This emulated framework analyzes the performance of the model. A novel and innovative early warning system is also created to send the warning on WhatsApp or email to several users in a single shot, when achieving certain goals such as file’s size limits and so on. In this research SVM, MCC, NLP and CNN machine learning algorithms are applied to detect students’ feelings and emotions to track the feedback via IoT enabled system.
Read MoreDoi: https://doi.org/10.54216/JISIoT.100206
Vol. 10 Issue. 2 PP. 36-75, (2023)
The primary objective of this research endeavour is to concentrate on the timely detection and prognostication of diabetes and Parkinson's disease through the utilisation of machine learning techniques, specifically the integration of Ant Colony Optimisation (ACO) with the XGBoost algorithm (ACXG). The healthcare issues presented by diabetes and Parkinson's disease underscore the criticality of early detection in order to facilitate effective intervention and enhance patient outcomes. The objective of this work is to establish a connection between the prediction of diabetes and the classification of Parkinson's disease, thereby developing a comprehensive model that improves the prognosis and prevention of these diseases. The project entails the collection and pre-processing of pertinent datasets, afterwards employing a range of classification approaches such as Logistic Regression, Support Vector Machine (SVM), Random Forest, and the innovative ACO-XGBoost model. The results of performance comparisons demonstrate that ACO-XGBoost has superior performance in contrast to conventional approaches. It achieves notable levels of accuracy, precision, recall, F1-score, and AUC, hence establishing its significance as a valuable tool for disease prediction. The incorporation of Ant Colony Optimisation (ACO) with XGBoost (ACXG) showcases the capacity to augment predictive precision and sensitivity, presenting notable progressions in healthcare methodologies. The present study makes a valuable contribution to the advancement of more accurate predictive models, ultimately enhancing the quality of patient care and public health outcomes.
Read MoreDoi: https://doi.org/10.54216/JISIoT.100207
Vol. 10 Issue. 2 PP. 76-89, (2023)
In the realm of Human-Computer Interaction (HCI), the importance of hands cannot be overstated. Hands serve as a fundamental means of communication, expression, and interaction in the physical world. In recent years, Augmented Reality (AR) has emerged as a next-generation technology that seamlessly merges the digital and physical worlds, providing transformative experiences across various domains. In this context, accurate hand pose and shape estimation plays a crucial role in enabling natural and intuitive interactions within AR environments. Augmented Reality, with its ability to overlay digital information onto the real world, has the potential to revolutionize how we interact with technology. From gaming and education to healthcare and industrial training, AR has opened up new possibilities for enhancing user experiences. This study proposes an innovative approach for hand pose and shape estimation in AR applications. The methodology commences with the utilization of a pre-trained Single Shot Multi-Box (SSD) model for hand detection and cropping. The cropped hand image is then transformed into the HSV color model, followed by applying histogram equalization on the value band. To precisely isolate the hand, specific bounds are set for each band of the HSV color space, generating a mask. To refine the mask and diminish noise, contouring techniques are applied to the mask, and gap-filling methods are employed. The resultant refined mask is then combined with the original cropped image through logical AND operations to accurately delineate the hand boundaries. This meticulous approach ensures robust hand detection even in complex scenes. To extract pertinent features, the detected hand undergoes two concurrent processes. Firstly, the Scale-Invariant Feature Transform (SIFT) algorithm identifies distinctive keypoints on the hand's outer surface. Simultaneously, a pre-trained lightweight Convolutional Neural Network (CNN), namely MobileNet, is employed to extract 3D hand landmarks, the hand's center (middle finger metacarpophalangeal joint), and handedness information. These extracted features, encompassing hand keypoints, landmarks, center, and handedness, are aggregated and compiled into a CSV file for further analysis. A Gated Recurrent Unit (GRU) is then employed to process the features, capturing intricate dependencies between them. The GRU model successfully predicts the 3D hand pose, achieving high accuracy even in dynamic scenarios. The evaluation results for the proposed model are very promising that the Mean Per Joint Position Error in 3D (MPJPE) is 0.0596 between the predicted pose and the ground truth hand landmarks, while the Percentage of Correct Keypoints (PCK) is 95%. Upon predicting the hand pose, a mesh representation is employed to reconstruct the 3D shape of the hand. This mesh provides a tangible representation of the hand's structure and orientation, enhancing the realism and usability of the AR application. By integrating sophisticated detection, feature extraction, and predictive modeling techniques, this method contributes to creating more immersive and intuitive AR experiences, thereby fostering the seamless fusion of the digital and physical worlds.
Read MoreDoi: https://doi.org/10.54216/JISIoT.100208
Vol. 10 Issue. 2 PP. 90-101, (2023)
The onset of digital education, propelled by the global COVID-19 crisis, has revolutionized the education sector, presenting unique difficulties, including the crucial task of preserving academic honesty. This study explores the possibilities of computer vision technologies, specifically face recognition and detection systems, to deter dishonest practices in online learning contexts. In this article we aim to construct efficacious strategies that leverage these technologies to track student actions in real-time and alert educators about possible cheating instances. This study presents two innovative models addressing cheating in online learning settings using cutting-edge computer vision techniques. Our initial model is an ensemble learning based face recognition system that blends the functionalities of three different deep learning (DL) structures: VGG, MobileNet, and DenseNet. This ensemble learning approach aims to offset the shortcomings of individual models while amplifying the overall effectiveness. The model’s efficiency will be gauged by juxtaposing it with other models and testing its performance against renowned benchmark datasets. Following this, we propose a second model designed for real-time face and cheating detection. This model integrates the FaceMesh model, facial landmarks analysis, and head pose estimation to identify possible cheating behaviors, such as significant shifts from a neutral or forward-facing head position. This model’s efficiency will be assessed through testing in simulated cheating scenarios and using authentic data from online learning contexts. Upon testing and validation, our proposed models have shown encouraging outcomes. The ensemble learning model outstripped individual models by attaining a remarkable accuracy rate of 91% through soft voting. Furthermore, the face detection system showcased sturdy abilities in recognizing faces under diverse conditions and accurately pinpointed potential cheating behaviors based on head pose estimation.
Read MoreDoi: https://doi.org/10.54216/JISIoT.100209
Vol. 10 Issue. 2 PP. 102-112, (2023)