The proposed IoT-based energy monitoring system is a powerful tool for managing and monitoring energy consumption in homes and small businesses. It is designed to measure the voltage and power consumed by appliances connected to it, providing users with real-time energy consumption data and a historical record of energy usage over time. Additionally, the system is equipped with features that enable remote access via a software application, allowing users to monitor their energy consumption and track their daily usage patterns to identify opportunities for energy savings. One of the most significant advantages of this system is its ability to turn off the appliances connected to it. Users can set up the system to turn off appliances when they are not in use, reducing energy consumption and lowering their electricity bills. The software application provides users with a comprehensive view of their energy consumption, allowing them to identify the appliances that consume the most energy and develop strategies to conserve energy. The application's data storage capabilities enable the safe and secure storage of energy consumption data, which can be accessed at any time. The appliance is a sophisticated energy monitoring system that operates by measuring the power and voltage consumption of appliances. It can be easily integrated into a home or small business network, and its data can be accessed remotely through the software application. With its powerful features, including real-time energy consumption monitoring, historical data tracking, and the ability to turn off connected appliances, this IoT-based energy monitoring system is an ideal choice for those looking to conserve energy and lower their energy bills.
In the pharmaceutical industry, accurate demand forecasting is crucial for the efficient management of manufacturing, acquiring, and distribution activities. This paper concentrates on utilising data mining techniques, namely the Decision Tree and Random Forest algorithms, to forecast the demand for pharmaceuticals associated with seasonal ailments that are anticipated to emerge in the upcoming months of 2024. The findings of the analysis demonstrate that the Random Forest algorithm exhibits superior performance in comparison to the Decision Tree algorithm. This is evidenced by the observation of lower mean Root Mean Square Error (RMSE) values of 80.53 and 97.10, respectively, which indicates that Random Forest outperforms Decision Tree. The present work's results offer significant contributions to the pharmaceutical sector by providing valuable insights that facilitate improved planning and resource allocation to address the variable demand patterns that are influenced by seasonal diseases.
In real-time embedded control (RTEC) systems, sensors collect data which is processed and sent to different control nodes. RTEC deployments have numerous applications in diverse verticals like industrial control, healthcare, and vehicular networks. In such cases, a trusted and verifiable control is required, particularly when the data is kept in a distributed manner, and is exchanged over open wireless channels. Thus, blockchain (BC) is a viable option to store the sensor data between RTEC systems, which maintains a trusted ledger of associated operations. Existing works have not focused on the integration of BC in RTEC systems. Motivated by the gap, the paper presents a systematic approach to integrating BC in RTEC ecosystems. We present a reference architecture and discuss the device registration, the hyperledger fabric set up, and the task offloading strategy between edge gateways and cloud nodes, and present the performance analysis of the architecture. The discussion of open issues and challenges also highlights the practical implications of the approach, emphasizing its importance for future deployments of real-time embedded control systems.
The fast growth of information and communication technologies (ICTs) has had a big impact and changed a lot of businesses. Wireless sensor networks (WSN) are used in many industries for mobility, scalability, reliability, smart monitoring, and management. Utilization of modern equipment in the agriculture industry is essentially required to boost people’s income and create a positive impact on their social lives. WSNs comprise various self-designed devices that gather extensive data from the environment. Numerous techniques have been developed recently to enhance WSN productivity in various industries. WSN has become one of the emerging innovations. In wireless sensor networks, the biggest problems are data loss, node failure, and the need to use more energy to make the sensor nodes last longer. To overcome these limitations, in this research, a novel Tri-Head Fuzzy C Multipath Routing (THFCMR) protocol is proposed for WSNs. The proposed THFCMR approach is designed by a tri-head static fuzzy C means clustering with hybrid energy-efficient distributed clustering (HEED) integrated with a hybrid energy-efficient multipath routing protocol (HEEMP) approach. This paper focused on reviewing the cluster base protocol’s usability and weaknesses, along with a proposal for a solution to enhance network consistency and data availability in WSNs.
Thеrе аrе numеrοus tіtlеs fοr сrурtο сurrеnсіеs. Mοst lіkеlу, уοu must hаvе hеаrd аbοut thе mοst wеll-knοwn сrурtο сurrеnсіеs, іnсludіng Bіtсοіn, Tеthеr, Еthеrеum. Сrурtο сurrеnсіеs аrе іnсrеаsіnglу рοрulаr аltеrnаtіvе fοr οnlіnе рауmеnts. А dіgіtаl сurrеnсу, οftеn knοwn аs а сrурtο сurrеnсу, іs а dіffеrеnt tуре οf рауmеnt sуstеm сrеаtеd utilising еnсrурtіοn mеthοds. Bу utіlіsіng еnсrурtіοn аlgοrіthms, сrурtο сurrеnсіеs mау sеrvе аs а vіrtuаl ассοuntіng sуstеm аnd аn еxсhаngе mеdіum. Tурісаllу, nο gοvеrnmеnt οr аnу οthеr central bοdу іssuеs οr сοntrοls сrурtο сurrеnсіеs. Thеу аrе mаnаgеd vіа рееr-tο-рееr nеtwοrks οfiсοmрutеrs runnіng οреn-sοurсе, frее sοftwаrе. Gеnеrаllу, аnуοnе whο wаnts tο jοіn thеm іn wеlсοmе tο dο sο.
In recent times, it has become increasingly popular to examine a wide range of environmental and earth data using remote sensing schemes supported by satellite images (SI). Due to the complex nature of spatial, spectral, and temporal characteristics of SI, it is quite difficult for automatic analysis to be performed as it requires specially designed algorithms. As part of the research proposal, an enhanced deep-learning scheme will be implemented in order to extract the waterbodies from the chosen SIs. The phases involved in this scheme includes; (i) the collection and resizing of images, (ii) Shannon's Entropy preprocessing, (iii) DeepLabV3+ extraction of waterbodies, (iv) the comparison of extracted sections with ground truth (GT) and the calculation of performance metrics, and (v) validation of the effectiveness of the implemented scheme. In the proposed work, the proposed multi-thresholding approach is combined with DeepLabV3+ in order to get the waterbodies to be extracted from the chosen test images. DeepLabV3 is demonstrated to have excellent segmentation performance when it is compared to UNet and SegNet. The experimental results of this scheme indicate that DeepLabV3+ results in higher Jaccard value (>89
High intraocular pressure causes the eye disease glaucoma, which can eventually result in complete blindness. On the other hand, early detection and treatment of glaucoma can prevent complete blindness in a patient. However, we regularly experience delays as a result of challenging glaucoma screening procedures and a shortage of human resources, which might raise the worldwide vision loss ratio. In the final stage, it is envisaged that a confined region comprising glaucoma lesions and associated classes will develop. To prove the technique’s viability, it was put to the test on a challenging dataset, specifically an online retinal fundus image database for glaucoma research (ORIGA). Due to the existing dearth of intelligence and security research on outdoor gantry cranes, a method based on the updated you-only-look-once (YOLO)v5 network for intelligent anti-intrusion detection is proposed. The first step is to offer a broad detection strategy. The YOLOv5 network’s goal is to retain speed while achieving the highest detection precision: Add multi-layer receptive fields and fine-grained modules to the backbone network to improve the performance of features. The training of YOLO V5 resulted in an accuracy of 92.5% by the end of the 100 th epoch. The high accuracy hence proves that the model was able to detect effectively.
The practical applications of blockchains can far supersede the widely known trading and cryptocurrency realm. If any service provider is looking for a consistent, immutable, and multitenancy-supported ledger, then blockchain is the promising solution. Nowadays, social engineering attacks are prevalent. And the attackers deceive cryptocurrency traders. This work investigates various ensemble learning, neural network, and machine learning algorithms for fraud detection and identifies the best decision-making algorithm. It is observed that Adaptive Boosting (AdaBoost) algorithm outperforms with an accuracy of 98.92%. Further, the fraud detection module is integrated with an application developed for cryptocurrency transactions. Before a new transaction is committed to blockchain, The fraud detection module intervenes and alerts the user. We have also designed a test bed of deployable Peer-to-Peer (P2P) network to simulate cryptocurrency transaction.
The rapid advancement of computer vision procedures has made the artificial raising of animals a more viable option for actual production settings. One such example is the critical need of enhancing the accuracy of day-age identification of birds in the poultry breeding industry. Within a time frame of 100 days, this article addresses the challenge of accurately categorising the age of hens. In actual application settings, when data volumes are large and device computing capabilities vary, it is crucial to make the most of the processing capabilities of edge computing devices without compromising data accuracy. An accurate deep learning-based model is proposed in this study for use in edge computing. This article takes a pre-trained DarkNet-53 model into account, performs model to make it suitable for low computing power circumstances, and runs a series of focused tests to ensure the model’s efficacy. Classification accuracy is enhanced by 3
In recent years, there has been an increase in interest in the field of education to use machine learning approaches to improve student achievement. Our study emphasizes the significance of including behavioral and psychological data in addition to standard academic data for a thorough understanding of student performance. We can discover important new information about the non-cognitive elements that have a big impact on learning outcomes by combining these many data sources. By enabling proactive interventions and customized instructional strategies that meet the specific needs of individual pupils, this multidimensional approach has the potential to revolutionize current educational practices. This study highlights the capability of machine learning methods to forecast students’ academic achievement based on psychological and behavioral information. In order to promote academic performance and improve educational outcomes for all children, educators and policymakers can use data analytics to make knowledgeable decisions and put evidence-based solutions into practice.
As technology advances daily, are advancing our lives into the digital sphere. The introduction of these cryptocurrencies aims to prevent the financial crisis. Due to its decentralized nature, high level of security, and restrictions on the number of coins that may be created, cryptocurrencies have attracted investors. Predicting the future price includes several limits and determinants because it involves capital. It varies according to market share. Using block chain technology and encryption, the transactions are encrypted from the beginning to end. Predicting prices to encourage consumers to invest during a specific period and earn a profit. They include a variety of elements, such as market analysis, sentiment analysis on Twitter, trading volume, and open and closing prices. Typical models that can be used to forecast bitcoin prices include regression techniques, neural networks, and support vector machines. Predictions of cryptocurrency prices based on their closing prices give investors additional insight into whether to wait until the closing period if prices are low for the entire day or to invest the following day. Using deep learning and bidirectional Long Short-Term Memory (LSTM) suggested this model to anticipate the price of digital currencies including Bitcoin, Litecoin, Ethereum, and Cardano. In this model, predictions are made using historical price statistics, and the graph is created by evaluating several performance indicators.
The relevance of automated recognition of human behaviors or actions stems from the breadth of its potential uses, which includes, but is not limited to, surveillance, robots, and personal health monitoring. Several computer vision-based approaches for identifying human activity in RGB and depth camera footage have emerged in recent years. Techniques including space-time trajectories, motion indoctrination, key pose extraction, tenancy patterns in 3D space, motion maps in depth, and skeleton joints are all part of the mix. These camera-based methods can only be used inside a constrained area and are vulnerable to changes in lighting and clutter in the backdrop. Although wearable inertial sensors offer a potential answer to these issues, they are not without drawbacks, including a reliance on the user’s knowledge of their precise location and orientation. Several sensing modalities are being used for reliable human action detection due to the complimentary nature of the data acquired from the sensors. This research therefore introduces a two-tiered hierarchical approach to activity recognition by employing a variety of wearable sensors. Dwarf mongoose optimization process is used to extract the handmade features and pick the best features (DMOA). It predicts the composite’s behavior by emulating how DMO searches for food. The DMO hive is divided into an alpha group, scouts, and babysitters. Every community has a different strategy to corner the food supply. In this study, we tested out a number of different methods for video categorization and action identification, including ConvLSTM, LRCN and C3D. The projected human action recognition (HAR) framework is evaluated using the UTD-MHAD dataset, which is a multimodal collection of 27 different human activities that is available to the public. The suggested feature selection model for HAR is trained and tested using a variety of classifiers. It has been shown experimentally that the suggested technique outperforms in terms of recognition accuracy.
The semiconductor industry is growing day by day and currently, all different countries are trying to invest in semiconductor development as these chips are very much useful in almost every type of business-like automobile, IT, electrical industries, healthcare, transportation, and many more industries. So, a system should be there to check the proper making of semiconductors such that there is less wastage of materials. These wafers are circular in shape and have sensors installed in them. These wafers are subjected to pass through various chemical and radiation processes which cause defects in them and moreover, these wafer’s size is also increased over time to fit in extra memory. Our objective is to make an ML model which will help industries to identify these defective wafers without physically examining them with the help of previously recorded data about defective wafers.
Radiological imaging of the chest (X-ray) is a cost-effective, widely accepted method of examining the lungs and abnormalities. During this research, a clinically significant Convolutional-Neural-Network (CNN) framework will be proposed for the examination of lung abnormalities. The proposed scheme aims to achieve better detection accuracy from the selected chest X-ray data. The following stages are included in these techniques: (i) CNN segmentation of the lung section, (ii) Deep-feature extraction utilizing selected CNN schemes, (iii) Handcrafted-feature extraction, (iv) Optimizing features using Firefly Algorithms, and (v) Binary classification and cross-validation of fivefold cross-validation. By implementing the pre-trained VGG-UNet, this framework is capable of extracting lungs sections from X-ray images. Using this lung segment, handcrafted features such as Local Binary Patterns (LBP) and Pyramid Histograms of Oriented Gradients (PHOG) are obtained. DFs and HFs are obtained using the FA, and then a serial concatenation is performed in order to obtain a hybrid feature vector. This feature vector is used to classify X-ray images into healthy and diseased groups. For examination in this study, X-ray images of disease classes, such as tuberculosis, COVID19, pneumonia, lung masses, and effusions, are considered. Based on the experimental results of this study, >98% of disease detection accuracy was confirmed using the proposed scheme combined with the SoftMax classifier.
Without a question, two of the maximum significant technologies to reach conventional IT in recent years are computing and big statistics analytics. The surprising convergence of the two technologies is yielding potent outcomes and advantages for enterprises. The delivery of IT services by so-called cloud firms and the relationship between enterprises and IT resources are already being affected by cloud computing. Recent developments in information and communication technology have made possible a new approach to data analysis known as “Big Data”. Nevertheless, the large quantity of computer resources needed for big data analysis means that many small and medium-sized businesses cannot afford to embrace big data technologies at this time. Affordances in business analytics, cloud computing data security are all part of the concept. Technique (KPCA-LDA-XGB) is used to conduct the empirical research. Using a structural equation model built using Partial Least Squares, this theory is experimentally evaluated with data from 316 businesses. Business analytics and the decision-making affordances of safety are positively moderated by data-driven ethos and IT business process integration. The findings of this research provide practical guidelines for organisations looking to advance their computing data safety organisation with the usage of analytics.
In the recent era, Machine Learning and Artificial Intelligence have come to a very great development point as we can use ML algorithms to predict the type of Erythemato-Squamous (Skin) diseases of the skin. In Dermatology, differential diagnosis of skin diseases is quite challenging in real life because most skin diseases share many histopathological features. And in this work, Psoriasis, Lichen Planus, Seborrheic Dermatitis, Chronic Dermatitis, Pityriasis Rosea, and Pityriasis Rubra Pilaris are among the skin illnesses for which eight different algorithm analytical comparison is done. Moreover, each classifier algorithm is discussed in detail with its pros and cons. The machine learning algorithms like Support Vector Machine, Decision tree, Random Forest, KNN, Naïve Bayes, Gradient Boosting, XGBoost, and Multilayer Perception have been proven to be successful in preserving state information through exact segmentation/classification. Random forest, Gradient Boosting, and XGBoost outperform all other methods and give an accuracy of 100% on the given ESD dataset. While Support Vector Machine gives the least accuracy of 72.97%. The paper also discusses the difficulties connected with skin disease segmentation or categorization. Furthermore, the study proposes future potential directions that include real-time analysis.
In the realm of unmanned aerial vehicles (UAVs), path planning is crucial. The objective is to create a safe, practical, and optimum flight route for UAVs to fly across unfamiliar terrain while avoiding obstacles. Because of their high computational complexity and inability to handle changing surroundings, traditional route planning algorithms confront hurdles in real-world applications. As a result, academics have created many sophisticated ways to handle UAV route planning challenges. This research provides a unique technique for intelligent UAV route planning. The program first divides the research region using the Delaunay Triangulation method. After obtaining the partitions, the coverage points are located, and the issue is framed as the Traveling Salesperson Problem (TSP). The Bat Optimization Algorithm (BOA) is used to find an optimum route for the TSP. The suggested technique seeks to find the best route for UAVs in dynamic situations. The suggested approach was verified by comparing it to other existing algorithms. The findings demonstrate that the suggested method can construct an optimum route that meets all requirements while avoiding obstacles. Moreover, the method may swiftly identify a new route as the environment changes. The study findings show that the suggested approach beats the other techniques in total computing time, saving roughly 0.03 s. Moreover, the suggested technique reduces the UAV’s route by around 0.02 cm, substantially improving existing algorithms.
Introduction:The neurodegenerative Disorder called Parkinson’s disease (PD) is incurable when it is indicated at an initial stage by weakening dopamine generating nerve cells. These cells are projected by the instrument of capturing DaTscan images. The qualitative analysis of DaTscan is carried out by the different deep learning algorithms. The present investigation deals with three facets of deep learning algorithms in diagnosing PD at an initial stage by analysing Volume Containing DaTscan Image Slices (VCDIS) for higher diagnostic accuracy. Methods: The contribution includes 3 facets. The first facet classifies the people suffering from PD from healthy individuals (HI) using transfer learning of conventional deep learning networks like Alex Net, Inception, Mobile Net, ResNet, Xception, VGG16, VGG19. Second facet fine tunes the layer of conventional networks carefully. Fine tuning is done by selecting random and optimal layers of conventional networks to tune the network to learn the features of VCDIS for better performance. Third facet hybrids the predictions of previous facets to accomplish highest proficiency in predicting PD at an early stage. Results: The results of the transfer learning facet are dictated as around 92.22% of predictive accuracy for the Inception V3 network. The diagnostic accuracy of optimal fine tuning is around 99.17% considerably. The best results of detecting EPD (Early Parkinson’s Disease) is achieved as 99.95% of accuracy for the Inception V3 network by the hybrid technique. Conclusion: The proposed hybrid technique runs remarkable performance as a supporting tool in analysing Parkinson’s disease which helps the neurologist in analysing the neurodisorders.
With the corona virus pandemic, social contact has been kept to a minimum, and education in schools can be carried out remotely. As a result of this, the concept of distance education has gained importance. In this study, natural language processing (NLP) and its effects on distance education are discussed, and by using NLP, a topic classification system is proposed. Classification is applied to text-based lesson questions in Turkish language. In this way, the questions asked by the students can be quickly directed to the teachers in the relevant specialty through a system to be designed, and the processes can be accelerated. In the data preparation phase, the real-world lesson questions were collected and converted from image to text using the EasyOCR library, and topic classification was performed on the data set using the Berturk model. Since the image-to-text method was used in the data set preparation phase, we encountered some noise in the data. To clean the data, different data preprocessing and cleaning techniques are applied. Finally, the training has been performed, and accuracy rates are presented.
Next-generation electronic voting systems have been made possible by the maturation of blockchain-based systems. Blockchain technology, the basis for electronic voting, might be used to improve online voting security. Nevertheless, due to its resource intensive nature, one of Blockchain’s key principles, Work, cannot be applied in the E-voting founded blockchain. As a consequence, the information included in a transaction block may be changed and its hash easily recalculated. In totalling, the Proof-of-Work is necessary to verify the block’s legitimacy. Hence, the digital signature may provide a means to address these concerns. Thus, the data is encrypted using an enhanced version of the Blowfish technique, which improves both security and efficiency. We also assess the security of the hybrid blockchain we propose and compare it to the security of the traditional blockchain via the discussion and analysis of attack execution. We learned new things about the safety, speed, of blockchain-based electronic voting schemes through our tests.
