The traditional methods of searching for flats are often inefficient and lack necessary features, making it difficult for both landlords and tenants to find what they are looking for. The “Sakanat” mobile application is designed to bridge this gap by offering a user-friendly platform for property listing, searching, and management. Landlords can easily upload details, photos, and set rental terms, while tenants can efficiently find suitable flats based on various criteria using advanced search filters. Rent payments can be made directly through the app using a secure payment gateway, and tenants can also access a variety of additional services such as maintenance companies and flat transfer services. Sakanat leverages Flutter for cross-platform app development and Firebase for backend services and data storage. Development will prioritize user-friendliness, performance, and security. By facilitating connections between landlords and tenants, simplifying rent collection, and offering valuable additional services, Sakanat has the potential to revolutionize the flat rental experience.
This study employs the Theory of Reasoned Action (TRA) to elucidate the relationship between key factors and the purchase intentions of Malaysian Generation Z consumers during online flash sales promotions. The research framework posits utilitarian value (UV) and hedonic value (HV) as components shaping consumer attitudes, while informational influence (II) and normative influence (NI) contribute to subjective norms. Additionally, perceived perishability (PP) and perceived scarcity (PS) are considered to capture the effects of flash sales strategies. Subsequently, six hypotheses are formulated. A quantitative approach is adopted, involving the distribution of questionnaires to 280 respondents within the 1625 age range, representative of Generation Z in Malaysia. Preliminary findings from this study indicate the viability of the research, with data analysis conducted using SmartPLS software. Results reveal significant impacts of UV, HV, II, PP and PS on the purchase intentions (PI) of Malaysian Generation Z during online flash sales promotions. However, NI demonstrates no significant effect. This study offers valuable insights for both researchers and marketers, informing future endeavors aimed at optimizing online flash sales promotions.
Road transport, utilizing roadways, is important for the movement of goods and people, posing a significant maintenance challenge due to the deterioration caused by poor materials and natural disasters. Cracks in asphalt roads are a major concern, impacting transportation quality, safety, and comfort. Identifying these cracks is crucial for maintaining high-quality roads. Nonetheless, the subjective nature of crack detection, time limitations, inaccuracies, and expenses restrict the extent of manual assessment by professionals. To address these issues, this study proposes YOLOv4, a deep learning solution adept at detecting road cracks. Utilizing Unmanned Aerial Vehicles (UAVs) or drones, the road pavement was captured in Perlis, Malaysia, at an altitude of 10 meters for optimal image acquisition. These images are then labelled and input into the YOLOv4 model for training, resulting in accurate crack detection and validation. Comparing the results with the YOLOv3, Faster RCNN, and SSD model, found that YOLOv4 significantly improves crack detection accuracy, especially when using UAV images, thus enhancing previous research outcomes.
The study focuses on Artificial Intelligence’s importance in developing green innovation. Artificial intelligence significantly enhances green innovation by enhancing productivity, accelerating environmentally friendly technological advancements, and facilitating better decision-making through energy optimisation, waste reduction, and smart infrastructure support. This research adopted a scoping review approach utilising the Scopus database as a source of documents. The study highlights several aspects, such as publication trend by years, publication source and context analysis. The reviews included documents published since 2021. The method section is derived from the PRISM-ScR checklist table. The highest publisher was the Business Strategy and The Environment journal, and the country with the highest publishing was China. In addition, the study recommends that more efforts should be exerted to increase companies’ awareness of the importance of green innovation and the feasibility of developing green products.
Cardiovascular disease (CVD) has become a significant concern, being a leading cause of illness and death in many parts of the world. India, being a developing nation, plays a significant role in the burden of cardiovascular disease due to its unique combination of factors such as demography, lifestyle, pollution, genetic tendency, and other influences. Traditional cardiovascular risk calculators may not always provide precise estimates for the Indian population due to variations in risk factors. Hence, there arose a necessity to incorporate machine learning algorithms in this domain. This research suggests an effective cardiovascular disease prediction system tailored specifically for the Indian population, utilizing a novel feature selection approach. This new feature selection method (NFS) is the integration of three different existing methods-Mean Gini decrease (MDG), Recursive feature elimination (RFE) and Mutual Information (MI) based selection. The novelty in the method is that the importance score of each feature was obtained from each traditional feature selection method. These scores were then normalized and added. The top 10 features were selected on the basis of total score. After selecting these 10 features Statistical technique of Pearson Correlation coefficient was used to check the correlation between these features which resulted in selection of optimal feature set of 8 features. Three different machine learning techniques were used to compare the performance of this NFS namely Random Forest (RF), Naïve Bayes (NB) and Logistic Regression (LR). Out of all, RF was found to be the best algorithm with accuracy of 97.5% and sensitivity of 98.2%. The preprocessing techniques used were missing value imputation by Proximity Matrix and Normalization by Min/Max method. K fold cross validation was used with value of K as 10 to avoid any overfitting.
Cancer remains a significant global cause of mortality, presenting substantial challenges for both medical practitioners and researchers. Recent data from a 2023 American Cancer Society study anticipate 238,340 new cases of lung cancer, with 127,070 resulting fatalities-constituting approximately $20 \%$ of all cancer-related deaths. Early detection plays a crucial role in improving survival rates across various cancer types, and precise classification based on medical images aids physicians in selecting optimal therapies to reduce cancer mortality. While extensive work has been conducted in lung cancer detection using Convolutional Neural Networks (CNN), challenges persist due to the complex structures present in CT scans. In addition, CNN models encounter challenges related to their functionality, such as choosing an ideal architecture, figuring out appropriate model parameters, and fine-tuning weights and biases. This paper introduces an innovative approach to lung cancer classification, utilizing deep learning techniques within a 1D framework based on pretrained neural network architectures. First, a 1D architecture is designed, and then the solution vector of the model is computed, providing a foundational understanding of the architectural framework. The study integrates three distinct pretrained classifiers-AlexNet, VGG-16, and VGG-19-into the 1D framework and trains them on a curated set of features crucial for accurate lung lesion classification. The proposed system produces promising results, with both AlexNet and VGG-19 achieving an impressive accuracy of $92 \%$, while VGG-16 outperforms them, reaching the highest accuracy at ${9 4. 6 7 \%}$. This research underscores the potential of leveraging pretrained neural networks within a 1D framework to enhance lung cancer diagnosis, contributing to more effective and timely interventions in the battle against this deadly disease. The findings confirm that the hybrid algorithm offers a reliable solution for the classification of lung cancer.
Retinal optic disk (OD) fundus imaging is a valuable tool for glaucoma diagnostics, while deep learning methods are often used to assist ophthalmologists in their clinical practice. Nonetheless, the performance of deep learning-assisted OD evaluation can be severely affected by the experiment variations and the use of small samples in network training. This study proposes using a particle swarm optimization (PSO)-optimized U-Net model to automatically segment OD in fundus photographs. The PSO method is used to fine-tune the training hyperparameters, i.e., learner, number of epochs, mini-batch size, and initial learning rate, to achieve good generalization performance even with a small training dataset. The findings showed a considerably good performance of the trained model with average overlap measures of 0.93-0.96 and mean performance evaluation metrics ranging between 0.89 and 0.99, comparable with the results of the state-of-the-art methods. This study concluded that the segmentation model trained using the proposed optimization framework could potentially be used for early glaucoma detection and improve healthcare delivery.
This paper proposes a new system to address challenge for reading accessibility through the development of a Character Recognition System (CRS) for blind people. This system has reading solution for aiding blind individuals by accessing printed letters. This paper introduces useful hardware for capturing images of printed letters using a manual printed machine and webcam, where a paper can be provided and moved in order to acquire its printed letters. So, a big dataset is collected for the printed English letters (a-z), it is named the Printed English Letters (PEL) dataset. After the acquisition, the input images are prepared, segmented and resized. Then, they are sent to a Deep Convolutional Neural Network (DCNN) for letters recognition. At the end, the recognized letter is suggested to be converted into audible speech, character-to-speech, in order to make the system more useful in helping blind persons. The main results were impressive and superior. That is, values of statistical metrics were recorded as 99.44%, 92.77%, 92.77%, 99.71% and 9 2.7 7% for the accuracy, precision, sensitivity, specificity and f1_score, respectively.
Sentiment analysis, sometimes referred to as opinion mining, is essential for understanding public opinion and attitudes toward various social topics and trends. This study aims to explore the effectiveness of machine learning (ML) models, namely support vector machine (SVM), long short-term memory (LSTM), and bidirectional encoder representations from transformers (BERT), in analyzing a dataset obtained from Kaggle, which contains 37,000 user reviews on the Instagram Threads app. After initial data cleaning and preprocessing, the dataset was partitioned into ${7 0 \%}$ for training and ${3 0 \%}$ for testing. Subsequently, the training set was used to create three datasets: a balanced dataset and two unbalanced datasets, one featuring $90 \%$ positive instances and the other featuring ${9 0 \%}$ negative instances. Subsequently, these datasets were used to train the three machine learning models mentioned above, resulting in nine different models. Evaluation metrics, including accuracy, precision, recall, and F1 score, were applied to assess model performance. The finetuned BERT model on the balanced dataset outperformed all the other models with an accuracy of $86 \%$, precision of $85 \%$, recall of $87 \%$, and F1-score of ${8 6 \%}$. Furthermore, these findings underscore the effectiveness of diverse ML techniques, particularly transformers, and the crucial role of data balancing in optimizing sentiment analysis tasks.
This paper presents an innovative approach to classifying the absorption performance of eco-friendly microwave absorbers in the L band using Multilayer Perceptron (MLP) networks. This project uses pyramidal absorbers coated with agricultural waste materials, such as empty palm oil bunches and coconut shells as carbon material to improve their absorption properties. The dataset consists of 87 absorption performance values of microwave absorbers obtained from experimental measurements using the NRL Arch Free. The objective of this study is to compare the effectiveness of three training algorithms which are Levenberg-Marquardt (LM), Resilient Backpropagation (RB) and Scale-Conjugate Gradient (SCG). The MLP network was trained using input parameters of frequency and absorption performance, and the performance of each algorithm was evaluated based on accuracy and mean-squared error (MSE). Results show that the LM algorithm with five hidden neurons achieved the highest training, validation and testing accuracy of 100% with the lowest MSE of 0.0455. These findings provide valuable insights for optimizing the design of microwave absorbers in the L band using eco-friendly materials.
Dysgraphia affects a person’s ability to write consistently and properly especially among school children. It is a challenging condition as it needs effective intervention to help the affected children succeed academically and socially. With the advancement in technology in artificial intelligence (AI), various methods and approaches have been developed using convolutional neural networks (CNN) model to overcome several limitations to assess dysgraphia symptoms. However, there are major concerns about the difficulties of getting large data of dysgraphia handwriting images for CNN attributes model. Thus, this study is aimed to develop dysgraphia handwriting recognition model based on augmentation method. In this study, image augmentation is addressed by creating new data by using rotation and brightness technique to generate a set of synthetic images. The augmented data is trained and tested using CNN classification model to classify four classes of dysgraphia handwriting. The results show a significant improvement with $77 \%$ accuracy using augmented as compared to without augmented data only $\mathbf{7 3 \%}$. This study indicated that augmentation method is significant for inclusion in CNN classification model particularly for dysgraphia potential risk recognition. This study is further recommended to implement intelligence-based augmentation method which can be incorporated into a computer-assisted dysgraphia screening system to provide a rapid, accurate, and unbiased dysgraphia detection.
In the field of access control locks, conventional methods often require users to download applications with Bluetooth connectivity, and biometric measures such as fingerprints or face recognition that introduce challenges like access to only authorized parties, limited device accessibility and intricate registration processes. To address these challenges, the research paper presents an innovative IoT-based Smart Door Lock System, aiming to redefine access control. Diverging from traditional approaches, this system eliminates the need for application downloads and simplifies the complexities tied to biometrics. The core innovation lies in the simplicity of a user-friendly 4-digit PIN for primary access, prioritizing ease of use. To overcome real-time access constraints, the system introduces One-time Password (OTP) and One Touch Access (OTA) modes. These modes eliminate the intricacies linked with traditional methods, ensuring immediate and hassle-free access for both authorized and unauthorized users.
The reliability and safety of large-scale solar photovoltaic (LSSPV) systems are crucial for effectively harnessing renewable energy. With the increasing adoption of solar energy in lightning-prone regions of Malaysia, understanding and enhancing protection mechanisms is essential. This study examined an induced current protection system for LSSPV using an early streamer emission (ESE) air terminal in Malaysia. Two systems, ESE and Franklin lightning rod types, were deployed in a 50 MWp PV power plant covering 260 acres, installed on the lightning arrester to ensure adequate protection. The Franklin rod system included 763 pieces and was constructed according to the Council of Engineer standards (Thailand). The ESE rod system comprised 68 pieces and was built following the NFC17102 standard (France). A 150 kA direct lightning impact on the PV power plant was then simulated using MATLAB/Simulink. The ESE lightning protection system (LPS) effectively protected and prevented damage from the lightning strike. Additionally, the Franklin rod system had more significant shading effects and higher installation costs (USD 10,026,800 vs. USD 8,026,800) compared to the ESE rod system. These results demonstrated the suitability of the ESE LPS for PV power plant implementation. The findings of this study could also help optimize lightning protection technology for large-scale PV power plants.
Headwater occurrences are defined by abrupt increases in water velocity and level as well as changes in water clarity. These events can seriously jeopardize infrastructure, human life, and ecosystems. Fatal accidents due to headwater phenomenon often occurs but no serious attention is given to develop headwater monitoring and early warning systems. The goal of this project is to provide a dependable Internet of Things (IoT) tool for headwater monitoring and early warning system. The system uses a pair of water level sensors to detect river water level and the flowrate of the river. The sensors were developed using mechanical micro limit switches as a sensor to monitor water levels. As for the cloud platform, Firebase and Google Sheets were used to store real-time data and automates it with Google Apps Script. LCD displays are used to display offsite warnings, and buzzers are used to deliver on-site alerts. This configuration improves safety by providing users with fast alerts and dependable data processing. It works well in scenarios involving high water levels since it gathers data accurately and notifies users in a timely manner.
Functional Electrical Stimulation (FES) devices represent a valuable rehabilitation intervention for patients with spinal cord injury (SCI). SCI can result in significant functional impairments, including the loss of movement in the lower body. Currently, closed-loop stimulation strategies are preferred over open-loop systems, as the latter’s trial-and-error approach can lead to premature muscle fatigue. However, while closed-loop systems offer promise, feedback controllers often exhibit inadequate performance when addressing nonlinear effects in knee muscles, such as stiffness, spasticity, and fatigue. To address these challenges, an adaptive feedback controller equipped with a system identification controller is crucial for accurately detecting these nonlinear parameters in knee muscles. This paper proposes the development of a digital system identification controller designed to detect nonlinear parameters from the knee extension model’s response. The system identification was implemented and simulated using hardware description language (HDL) Verilog code, with its performance evaluated through HDL co-simulation in MATLAB Simulink. The simulation results indicate that all nonlinear effects including stiffness, spasticity and fatigue are identifiable once the rate-of-change response stabilizes at 4 s.
This study investigated the influence of artificial intelligence (AI) on university decision-making for university students in Oman. Eight bachelor students were selected from various academic programmes to interview them to explore their experiences towards using AI in their academic related decision-making. The research data was analysed with thematic analysis using NVivo software. Although artificial intelligence supports the students’ decision-making processes with data-driven based decisions, the findings also shed light on concerns in relation to the risk of decline in personal creativity, critical thinking, and social relationships. The integration of artificial intelligence in education, therefore, should be more balanced to maximise the benefits of technology with essential human skills.
In this study, we present an innovative pharmacokinetic modeling that combines fractional order kinetics with deep learning techniques to improve the prediction accuracy of drug concentration distribution in biological systems. Traditional pharmacokinetic models rely on ordinary differential equations, which often fail to accurately predict complex drug behaviors. To address these limitations, we develop a fractional-order pharmacokinetic model that can more accurately represent the memory and genetic properties of biological systems. To complement this, we combine fractional-order pharmacokinetics with the predictive capabilities of Long Short-Term Memory (LSTM) models in recurrent neural networks (RNNs). Synthetic datasets were first generated by selecting appropriate pharmacokinetic parameters to simulate a range of drug behavior scenarios. The dataset was then used to train and optimize the LSTM model, aiming to improve the prediction of the fractional-order model and thus achieve higher accuracy. Compared to traditional models, our results show significantly better prediction accuracy, with lower root mean square error (RMSE) and higher coefficient of determination (R 2 ) values across time. This study not only highlights the potential of combining fractional-order kinetics with deep learning to improve pharmacokinetic models, but also opens avenues for more personalized and accurate drug therapy planning.
Generative AI models have witnessed remarkable advancements, blurring the lines between human creativity and machine generation. This paper concisely reviews the current Generative AI landscape, exploring its diverse applications across various domains. We delve into the capabilities of these models, from creating images and music to generating creative text formats. Furthermore, the paper examines the real-world applications of Generative AI, highlighting its potential to revolutionize industries like design, marketing, education, and scientific discovery. However, while existing research extensively explores specific aspects of Generative AI, an analysis of the technology’s landscape, encompassing its capabilities, applications in content creation, and regulatory considerations, remains limited. This paper strives to bridge this gap by delivering a more holistic landscape of GenAI. Our analysis of the GenAI landscape pinpointed user behavior research and responsible development practices as key to user-centric AI creation. Through this study, we aim to stimulate discussion and collaboration between researchers, developers, and policymakers to ensure this powerful technology is harnessed responsibly for the benefit of industry and society.
This paper analyzes the power consumption and maximum demand to evaluate the effect of the power factor and forecast the value of the power consumption and maximum demand using support vector machine (SVM) regression in MATLAB software. The data on power consumption and maximum demand are collected from the previous 12-month electricity bills for the year 2019 until 2022 at Universiti Teknologi Mara (UiTM) campuses: Pulau Pinang, Arau, and Tapah. The graph of the maximum demand, power consumption, and power factor has been plotted to show the relationship between these three variables for the year 2022. From the graph, it was found that when the power factor is low, more current is needed to supply the same amount of power to a load. This means that more power is being drawn from the electrical supply system, which increases the power consumption. The power factor also shows the efficiency of the electricity at the UiTM. The quadratic SVM in regression learning is used to forecast the value of power consumption and maximum demand in MATLAB software without or with the use of the Principal Component Analysis (PCA). Therefore, it can show which one is better to use to make a prediction. This study successfully achieved the objective of analyzing power consumption, maximum demand, and power factor and forecasting the value of power consumption and maximum demand.
The AC-AC converter is a device commonly used in the industrial sector today due to its ability to convert AC power at high frequencies. However, despite its widespread use, the AC-AC converter faces challenges in providing a smooth output. A major issue is the high Total Harmonic Distortion in voltage (THDv) at the output, which negatively affects system performance. This study proposes an improved solution by replacing the Silicon-Controlled Rectifiers (SCRs) with two bidirectional exclusive switches. These switches offer better controllability, faster response times, and lower switching losses compared to SCRs. When integrated into the AC-AC converter topology, they lead to a significant reduction in THDv, achieving 0.96% for a duty cycle of 30% under single-stage DC modulation, and 2.28% for the same duty cycle under double-stage DC modulation. Both modifications result in a smoother output waveform.
