MMU Press (Multimedia University)
Not a member yet
714 research outputs found
Sort by
Cuffless Non-invasive Blood Pressure Measurement Using CNN-LSTM Model: A Correlation Study
No full textCardiovascular disease is a major concern for people all around the world and still remains as the main cause of death worldwide. Blood pressure has been identified as the most important risk factor. Having the ability to acquire continuous monitoring on this biological parameter plays a significant role in reducing the risk of getting cardiac disease. Many studies conducted utilize two biosignals and features manually extracted from signals as input to the model. However, these methods increase the computational complexity in the pre-processing stage as it involves signal synchronization, and the model performance is highly dependent on the selection of features. The main objective of this study is to build a hybrid convolutional neural network combined with Long-Short Term Memory (CNN-LSTM) model to estimate blood pressure from PPG signals, which eliminates the need for manual feature extraction. Correlation study is performed to evaluate the performance of the model, and it gives a direct visualization of the model’s performance in percentage. This research compared the correlation performance between MIMIC-II dataset, UKM dataset, and PPG-BP dataset using the CNN-LSTM model to estimate blood pressure from PPG signals. The results show that the UKM dataset performs the best, having the highest overall correlation at 0.53 for systolic blood pressure, and 0.29 for diastolic blood pressure. The model trained with this dataset is suitable to estimate systolic blood pressure ranging from 141 to 150mmHg, and diastolic blood pressure ranging 81 to 90 mmHg. In conclusion, among the three datasets, UKM dataset is the most suitable dataset to be used as the input of the CNN-LSTM model to perform cuffless blood pressure measurement with PPG signals.
Manuscript received: 16 June 2023 | Revised: 18 July 2023 | Accepted: 27 August 2023 | Published: 30 September 202
Voice Controlled Home Automation System Design
No full textThe primary objective of this home automation study is to facilitate the implementation of a voice-controlled capabilities, primarily designed to assist individuals with disabilities or seniors. The system presented in this research enables wireless control of various household devices, such as lights, fans, and any electrical appliances through a dedicated mobile application. This voice-controlled home automation system leverages an android smartphone and a microcontroller using an android application to manage household appliances. The system can be accessed by user name and password. This system is designed with three main control interfaces such as Bluetooth connectivity, voice recognition, and manual control switching. The google cloud speech API which can convert spoken words into text is utilized for voice recognition. This generated text is then transmitted to the designated slave device, facilitating home automation through Bluetooth communication. Additionally, the system provides a manual control switch for user convenience. To control high-voltage appliances safely, an enhanced microcontroller is employed, incorporating a relay circuit for ON/OFF functionality. The developed prototype, encompassing both hardware and software components, has undergone comprehensive testing, validating its security features and compatibility with various home appliances. This innovative home automation system not only offers enhanced convenience but also prioritizes security, providing an attractive alternative to commercial solutions.
Manuscript received: 20 May 2023 | Revised: 20 August 2023 | Accepted: 1 September 2023 | Published: 30 September 202
Improvement of Pulse Shape by Reducing Ripple Using Coaxial Cable
No full textEven though pulse shape in rectangular form is desirable for most applications, pulse generated from coaxial cable has ripple and is not in desired rectangular shape. This study is to improve the pulse shape generated from coaxial cable to be rectangular shape. Coaxial cables with difference dielectric materials are studied for generation 100 ns pulse. The study has been carried out by state space representation method and MATLAB simulation for pulse forming network of Type-B which has same characteristics of coaxial cable. Ripple can be reduced to 2% to 3% increasing the inductance values next load higher than characteristics inductance of coaxial cable.
Manuscript Received: 14 February 2023, Accepted: 28 March 2023, Published: 15 September 2023, ORCiD: 0009-0002-6869-820
TracWork: An On-Field Employee Tracking System
No full textThe internet has transformed the world into a global village, benefiting our society as a whole and empowering people in a variety of ways. Many mobile applications are becoming a part of people's daily lives and assisting them in their jobs or daily routines, thanks in part to the phenomenal growth of Internet usage over the last 21 years. Previous research has found a scarcity of high-quality apps that cover all bases. This project's primary goals are to combine fragmented market systems into a product capable of performing functions such as tracking an employee's on-field movement using GPS; assisting employees in navigating to their next destination; maintaining and improving productivity levels using indicators such as battery status; current and past location; and so on. We highlight previous work and how we learned to extract a model that harmonises current systems while also improving quality of life in this study. We investigated numerous approaches, methods, and procedures before applying them to the development of the system.
Manuscript Received: 9 June 2023, Accepted: 18 July 2023, Published: 15 September 2023, ORCiD: 0000-0001-5038-800
Joseph Shine v Union of India: Farewell to a Victorian-Era Adultery Law
No full textThis case commentary critically analyses the rationale behind the decision made by the Indian Supreme Court in the case of Joseph Shine v Union of India [2018] Indlaw 899 (SC). The constitutionality of section 497 of the Indian Penal Code and section 198(2) of the Code of Criminal Procedure, which criminalise adultery, was challenged in this case. Being well aware that this case was a call made due to societal changes, the Supreme Court was prepared to adopt a liberal interpretation of the Indian Constitution. However, it had to face the sea of precedents flowing in the opposite direction of the societal changes. The Supreme Court, in dealing with these archaic provisions had carefully scrutinized Articles 14, 15 and 21 of the Indian Constitution to declare that the impugned provisions have long outlived their purpose and do not fit within today’s constitutional morality. This case is definitely one of the significant decisions made in the history of Indian law as it portrayed the Supreme Court’s bold move in finally bidding farewell to a Victorian-era law. As a result, adultery is no longer a crime in today’s India and this decision is the reason behind it
The Need for Artificial Intelligence in Solving Unsolved Criminal Cases and Sentencing in Malaysia
No full textAs humans, it is common for judges to give wrong verdicts when making decisions, especially in criminal cases. As such, those who feel that they have been wronged by the courts will thus appeal against the decisions. Due to the sheer volume of appeals, it has resulted in a backlog of cases. However, there is no one solution to solve the problem other than calling the judicial officers to improve themselves with legal knowledge before the real use of Artificial Intelligence in legal policy. In the current digital era, it is believed that Artificial Intelligence can accelerate and automate the review of potential evidence in identifying the most relevant and accurate evidence. With the help of Artificial Intelligence, it will reduce court unsolved cases. Countries such as the United States of America, Colombia, and China have started implementing Artificial Intelligence in their respective judicial systems. Yet Singapore’s criminal courts have no plan to use Artificial Intelligence in sentencing. Therefore, it has raised questions like should Malaysia’s judicial response to the use of Artificial Intelligence in cracking those backlog criminal cases and how far could it go in helping the judges. This paper seeks to highlight the issues
Electronic Waste Taking Over the Globe: An Overview of the Law in Malaysia, India, China and the United Kingdom
No full textThe rapid growth of the global market for electrical and electronic equipment (EEE) has resulted in an alarming increase in electronic waste (e-waste). E-waste, which comprises discarded electrical and electronic devices, poses severe environmental and health risks due to improper disposal and treatment. Many electronic devices contain hazardous materials that can leach into the environment, contaminating soil, water, and air. This paper explores the e-waste situation in Malaysia, India, China, and the United Kingdom, along with the laws and regulations governing e-waste in these countries respectively. While some countries have implemented adequate laws, compliance, and enforcement remain significant challenges. Encouraging extended producer responsibility (EPR), ensuring compliance with laws, and investing in recycling infrastructure are proposed as effective strategies for managing e-waste. To tackle this global problem, governments, industries, and individuals must collaborate, raise public awareness, and promote sustainable practices to build a circular economy for electronic products. By taking collective action, we can protect the environment and create a cleaner and greener future
Robust Image Watermarking With Quaternion Fractional-Order Polar Harmonic-Fourier Moments Based On Wavelet Transformation: Resistance Against Rotation Attacks
No full textThis study presents a zero-watermarking algorithm that can resist rotation attacks. The algorithm uses quaternion fractional-order polar harmonic-Fourier moments (QFr-PHFMs) based on wavelet-transformation. First, the wavelet-transformation is applied to each component of the host image, which is in RGB three-channel color. The low-frequency sub-bands of each component are then extracted and represented using quaternion algebra. Multiple QFr-PHFMs are calculated, and the invariants of the QFr-PHFMs are utilized to establish the watermark system. The watermark extraction process is also simplified. The detection of the image requires a two-level wavelet transformation, followed by the calculation of multiple invariant moments of the low-frequency sub-image. The experimental results are shown and compared with similar methods. Simulations show that this method can produce high-quality visual effects and withstand noise, filtering, JPEG compression, and cropping attacks
A Multi-Scale Feature Attention Image Recognition Algorithm
No full textThe success of image classification using small samples is contingent on neural network models' capability to derive image representations from the data. A proposed solution is a small-sample image classification system that leverages attention mechanisms and meta-learning to capture more comprehensive image information. Due to its ability to efficiently suppress irrelevant characteristics and accentuate pertinent ones, this technique may extract more robust multiscale features and enhance classification performance through meta-learning.In this paper, the effectiveness of the multi-scale attention network is verified on two datasets, namely, Mini-ImageNet and Tiered-ImageNet, and the accuracy of the method is 58.54% for 5-way 1shot and 74.76% for 5-way 5shot on the Mini-ImageNet dataset. In the dataset of the Tiered-ImageNet,the accuracy of 5-way 1-shot and 5-way 5-shot increased to 59.74% and 78.65%, respectively. The experimental results show that the multi-scale sub-attention can pay more attention to the global information of the image than the single-scale attention network, and significantly improve the accuracy of small-sample image classification
Predicting Travel Insurance Purchases in an Insurance Firm Through Machine Learning Methods After COVID-19
No full textTravel insurance serves as a crucial financial safeguard, offering coverage against unforeseen expenses and losses incurred during travel. With the advent of the proliferation of insurance types and the amplified demand for Covid-related coverage, insurance companies face the imperative task of accurately predicting customers’ likelihood to purchase insurance. This can assist the insurance providers in focusing on the most lucrative clients and boosting sales. By employing advanced machine learning techniques, this study aims to forecast the consumer segments most inclined to acquire travel insurance, allowing targeted strategies to be developed. A comprehensive analysis was carried out on a Kaggle dataset comprising prior clients of a travel insurance firm utilizing the K-Nearest Neighbors (KNN), Decision Tree Classifier (DT), Support Vector Machines (SVM), Naïve Bayes (NB), Logistic Regression (LR), and Random Forest (RF) models. Extensive data cleaning was done before model building. Performance evaluation was then based on accuracy, F1 score, and the Area Under Curve (AUC) with Receiver Operating Characteristics (ROC) curve. Inexplicably, KNN outperformed other models, achieving an accuracy of 0.81, precision of 0.82, recall of 0.82, F1 score of 0.80, and an AUC of 0.78. The findings of this study are a valuable guide for deploying machine learning algorithms in predicting travel insurance purchases, thus empowering insurance companies to target the most lucrative clientele and bolster revenue generation