Indonesian Journal of Electrical Engineering and Informatics (IJEEI)
Not a member yet
776 research outputs found
Sort by
A Cost Sensitive SVM and Neural Network Ensemble Model for Breast Cancer Classification
Get PDFBreast Cancer has surpassed all categories of cancer in incidence and is the most prevalent form of cancer in women worldwide. The global incidence rate is seen to be highest in the country of Belgium as per statistics of WHO. In the case of developing countries specifically, India, it has overtaken other cancers and stands first in incidence and mortality. Major factors identified as impacting the prognosis and survival in the country is chiefly the late diagnosis of the disease and diverse situations prevailing in different parts of the country including lack of diagnostic facilities, lack of awareness, fear of undergoing existing procedures and so on. This is also true for many other countries in the world. Early diagnosis is a vital factor for survival. The implementation of machine learning techniques in cancer prediction, diagnosis and classification can assist medical practitioners as a supplementary diagnostic tool. In this work, an ensemble model of a polynomial kernel-based Support Vector machines and Gradient Descent with Momentum Back Propagation Artificial Neural Networks for Breast Cancer Classification is proposed. Feature selection is applied using Genetic Search for identifying the best feature set and data sampling techniques such as combination of oversampling and undersampling and cost senstivke learning are applied on the individual Neural Network and Support Vector Machine classifiers to deal with issues related with class imbalance. The ensemble model is seen to show superior performance in comparison with other models producing an accuracy of 99.12%
Neuro-Fuzzy Combination for Reactive Mobile Robot Navigation: A Survey
Get PDFAutonomous navigation of mobile robots is a fruitful research area because of the diversity of methods adopted by artificial intelligence. Recently, several works have generally surveyed the methods adopted to solve the path-planning problem of mobile robots. But in this paper, we focus on methods that combine neuro-fuzzy techniques to solve the reactive navigation problem of mobile robots in a previously unknown environment. Based on information sensed locally by an onboard system, these methods aim to design controllers capable of leading a robot to a target and avoiding obstacles encountered in a workspace. Thus, this study explores the neuro-fuzzy methods that have shown their effectiveness in reactive mobile robot navigation to analyze their architectures and discuss the algorithms and metaheuristics adopted in the learning phase
Influence of Renewable Energy Sources on Day Ahead Optimal Power Flow Based on Meteorological Data Forecast Using Machine Learning: A case study of Johor Province
Get PDFThis article investigates a day ahead optimal power flow considering the intermittent nature of renewable energy sources that involved with weather conditions. The article integrates the machine learning into power system operation to predict precisely day ahead meteorological data (wind speed, temperature and solar irradiance) that influence directly on the calculations of generated power of wind turbines and solar photovoltaic generators. Consequently, the power generation schedulers can make appropriate decisions for the next 24 hours. The proposed research uses conventional IEEE -30-bus as a test system running in Johor province that selected as a test location. algorithm designed in Matlab is utilized to accomplish the day ahead optimal power flow. The obtained results show that the true and predicted values of meteorological data are similar significantly and thus, these predicted values demonstrate the feasibility of the presented prediction in performing the day ahead optimal power flow. Economically, the obtained results reveal that the predicted fuel cost considering wind turbines and solar photovoltaic generators is reduced to 645.34 USD/h as compared to 802.28 USD/h of the fuel cost without considering renewable energy sources. Environmentally, CO2 emission is reduced to 340.9 kg/h as compared to 419.37 kg/h of the conventional system. To validate the competency of the whale optimization, the OPF for the conventional system is investigated by other 2 metaheuristic optimization techniques to attain statistical metrics for comparative analysis
Enhanced Emotion Recognition in Videos: A Convolutional Neural Network Strategy for Human Facial Expression Detection and Classification
Get PDFThe human face is essential in conveying emotions, as facial expressions serve as effective, natural, and universal indicators of emotional states. Automated emotion recognition has garnered increasing interest due to its potential applications in various fields, such as human-computer interaction, machine learning, robotic control, and driver emotional state monitoring. With artificial intelligence and computational power advancements, visual emotion recognition has become a prominent research area. Despite extensive research employing machine learning algorithms like convolutional neural networks (CNN), challenges remain concerning input data processing, emotion classification scope, data size, optimal CNN configurations, and performance evaluation. To address these issues, we propose a comprehensive CNN-based model for real-time detection and classification of five primary emotions: anger, happiness, neutrality, sadness, and surprise. We employ the Amsterdam Dynamic Facial Expression Set – Bath Intensity Variations (ADFES-BIV) video dataset, extracting image frames from the video samples. Image processing techniques such as histogram equalization, color conversion, cropping, and resizing are applied to the frames before labeling. The Viola-Jones algorithm is then used for face detection on the processed grayscale images. We develop and train a CNN on the processed image data, implementing dropout, batch normalization, and L2 regularization to reduce overfitting. The ideal hyperparameters are determined through trial and error, and the model's performance is evaluated. The proposed model achieves a recognition accuracy of 99.38%, with the confusion matrix, recall, precision, F1 score, and processing time further quantifying its performance characteristics. The model's generalization performance is assessed using images from the Warsaw Set of Emotional Facial Expression Pictures (WSEFEP) and Extended Cohn-Kanade Database (CK+) datasets. The results demonstrate the efficiency and usability of our proposed approach, contributing valuable insights into real-time visual emotion recognition
Prediction of Power Consumption Utilization in a Cloud Computing Data Centre using Kalman Filter parameters with Genetic Algorithm
Get PDFData Centre (DC) has become a critical computing infrastructure that is essential to modern society by providing services such as cloud computing, Internet of Things (IoT) and big data. However, the cost of maintaining DC continues to rise as the demand for information technology services increase and this situation is further exacerbated in a country like Nigeria where there is highly unstable power supply from the national grid. The optimization of energy consumption in cloud computing DC using Genetic Algorithm (GA) to minimize the consumption of energy thereby extending network lifespan was one of the techniques used for optimization of power consumption. But the optimization was carried out with the assumption that all the parts of the modular server that are not carrying traffic is on idle mode and not completely off which consumes extra power compare to when it is completely off. Therefore, this work proposed optimization of power consumption utilization in a cloud computing DC using Kalman Filter (KF) with GA. Historical consumption trend and network traffic is analyzed to reduce the amount spent on power with assumption that servers in the DC operate as modular units which can be powered separately as required, in contrast to keeping entire servers always powered. Data from five different servers were collected from MTN Abuja DC in Nigeria. The servers were named BSC 13, BSC 14, BSC 15, RNC 05 and RNC 06. These consist of data recorded for two year - 5th January to 30th December 2019 as well as 5th January to 31st December 2020. The GA optimizer is used to obtain the best possible values for the Kalman Filter (KF) parameters. Then, the KF model is used to predict the future power consumption value on hourly basis for each day of the week. The proposed model gives low power consumption with accurate prediction when compared with the existing models.
Day Ahead Energy Consumption Forecasting Through Time-Series Neural Network
Get PDFDemand response management through appliance scheduling can effectively decrease electricity bills. Similarly, it can decrease the peak demand of both consumers and the utility grid. However, prior knowledge of the load profile of the consumer is required for effective appliance scheduling. This work developed a novel time-series forecasting model within the shallow neural network framework to predict the load curve for optimal planning of demand response by a nonlinear autoregressive network with exogenous input network. The algorithms such as Levenberg-Marquard, scaled conjugate gradient, and Bayesian regularization was applied to train the model. The results were compared with conventional seasonal autoregressive integrated moving average statistical model and long short-term memory network. The results indicated that the optimum methodology is a nonlinear autoregressive network with exogenous inputs using the Bayesian regularisation algorithm, which has the lowest MSE value in the training and testing phases of 0.0031 and 0.0029, respectively. It is practical to continue designing artificial neural networks to analyze hourly load consumption in the context of the positive outcomes acquired
Generic Solution Architecture Design of Regulatory Technology (RegTech)
Get PDFRegulatory Technology, or RegTech, uses new technology that assists the financial industry, such as FinTech and banks, in meeting regulatory compliance. RegTech automates various regulatory compliance activities that were previously manual, such as regulatory interpretation and regulatory reporting, amidst the challenges of the increasing volume of regulations and operational data. Some cutting-edge technologies discovered at RegTech include big data analytics, artificial intelligence, machine learning, robotic process automation, and cloud computing. Although very dominant in the financial industry, RegTech solutions have the potential to be applied in other regulated industries besides finance. Several studies have explored the potential for applying RegTech in industries other than finance, such as charitable organizations, real estate marketplace, pharmaceuticals, and healthcare. Therefore, this study aims to design a generic RegTech solution architecture so that it can be adopted and applied in various regulated industries achieve regulatory compliance more efficiently. Based on the evaluation results, the proposed architecture can be applied in an industrial environment other than financial to be considered generic. Furthermore, an evaluation of the comparison of regulatory compliance business processes without and by implementing RegTech can produce a time efficiency of 95.16%. These results show that RegTech solutions can achieve regulatory compliance more efficiently
Design of 28 GHz Microstrip Patch Antenna for Wireless Applications
Get PDFThis research paper presents a 28 GHz microstrip patch antenna design and analysis for mobile phone applications. Fr-4 (lossy) material, whose dielectric permittivity is 4.3 and loss tangent is 0.025, has been used as a substrate material for the antenna. Besides, copper annealed has been used in the ground, and the thickness of the patch is 0.035. CST software creates and simulates the complete antenna. Among the results obtained from the simulation, return loss, VSWR, directivity gain, and bandwidth are -24.507 dB, 1.126, 7.19 dBi, and 1.352 GHz, respectively. The main objective of this proposed antenna is to achieve an excellent VSWR value by reducing the return loss, increasing the antenna's directivity gain, and improving the bandwidth. As a result, this proposed design can be used on super-high-frequency devices (mobile phones) in the future. The fact that the results obtained from the suggested antenna design are superior to those reported in papers published in the past hints that the research has achieved increased performance compared to studies already conducted in the field
MIMO System Capacity based on Multiple fading channels and varying numbers of antennas including the comparison of MIMO-NOMA and MIMO-OMA
Get PDFMultiput-Input, Multiple-Output (MIMO) systems have greatly improved data speed, connection stability, and spectrum efficiency, revolutionizing wireless communication. In cases where numerous fading channels coexist and the number of antennas at both the transmitter and receiver fluctuates, this paper examined the functionality of MIMO systems. In actual wireless settings, fading channels frequently occur, resulting in time-varying and spatially correlated channel characteristics. In this paper, we used a thorough analysis to investigate the capability of MIMO systems under some difficult circumstances. We considered circumstances with various numbers of antennas, channel correlation, and fading statistics, ranging from Single-Input, Single-Output (SISO) to MIMO setups. We investigated the impact of geographic diversity, Rayleigh, Rician, Nakagami fading models, and correlated fading channels on system performance. The trade-offs between the quantity of antennas, channel correlation, and capacity in MIMO systems are well illustrated by our findings. We presented our evidence in this paper to show that MIMO-NOMA is solely superior to MIMO-OMA in terms of total channel capacity, except in scenarios where communication is limited to one individual, with a power disparity for which MIMO-NOMA can achieve strictly larger rate pairs than MIMO-OMA. This study also explored the outage probability (OP) performance of MIMO-NOMA and MIMO-OMA systems in a massive MIMO communication scenario, including different fading channels. Based on these findings, we demonstrated that MIMO-NOMA can achieve a higher total ergodic capacity than MIMO-OMA
AIoTST-CR : AIoT Based Soil Testing and Crop Recommendation to Improve Yield
Get PDFAgriculture is a backbone of any country. Farmers need to test the soil fertility and nutrients present in the soil for proper growth of the crops. In traditional system, the farmers collect soil sample and submit to soil testing labs for testing the soil nutrients and get the soil test reports manually. Farmers based on his experience and the season; decide which crop to be taken in the farm. Based on soil testing reports farmers decide which fertilizers required for the proper growth of the crop. This process is time consuming and human efforts are required and hence crop yield is affected. The recent technologies in cloud storage, wireless sensors, and AI based algorithms are very instrumental in decision making process of crop growth life cycle. Farmers can make use of mechanical automation tools for seeding, watering, supplying fertilizers, crop cutting etc. for proper growth of the crop. However, to observe the crop growth during the entire life cycle of crop farmer has to take lot of efforts to check need of water, any problem of disease to the crop, any specific fertilizers required or not and whether there is a need of harvesting. A proper decision support system is needed for helping the farmers in all such activities. Such support can be provided to a farmer so that he will be well updated about the growth of his crop in the farm. To reduce the human efforts and improve the crop yield, Artificial Intelligence and IOT based soil testing and Crop Recommendation system (AIoTST-CR) is designed and developed. AIoT based handheld soil testing system has pH, Nitrogen, Phosphorous, Potassium and Soil moisture sensing capability. A mobile application is developed to fetch the sensed data from AIoT system. A historical data is inputted to give training to ML models. Machine learning algorithm is used to predict and recommend the crop to be taken. The results show AIoTST-CR which is AIoT based soil testing and crop recommendation system provides effortless and accurate recommendations of crop. Our findings indicate that AIoT based system provides high accuracy, which outperforms existing commonly, used machine learning based crop recommendation systems