UTAR Institutional Repository (Universiti Tunku Abdul Rahman)
Not a member yet
6132 research outputs found
Sort by
Study on interactive effect of heat treatment and enzymatic activities in mature coconut water upon storage
Get PDFMature coconut water (MCW), though nutritionally valuable, is often discarded during coconut milk production. Factors such as heat treatment, sugar addition and enzymatic activity could lead to browning of MCW during processing and storage. Phenolic compounds in coconut water serve as substrate for enzymatic browning. Chitin, a renewable biopolymer has been widely applied in wastewater treatment and white wine production to remove phenolic compounds. However, research on its application in coconut water is limited. This study examined the effects of heat treatments (water bath, W; pressure cooking, P) and their combinations with chitin (CW, CP) on enzymatic browning in MCW. Changes in total phenolic content (TPC), browning index (BI), polyphenol oxidase (PPO) activity and peroxidase (POD) activity were monitored over 12 days. POD were absent in all samples, whereas PPO activity was highest in untreated MCW (control) and progressively decreased in W (64.00 ± 0.00 U/mL), CW (40.00 ± 0.00 U/mL) and was undetectable in P and CP (0.00 ± 0.00 U/mL) by Day 12. W and CW did not inhibit the browning (a progressive increase of BI from Day 0 to Day 12 storage). In contrast, P and CP treatments significantly suppressed browning in MCW, with no significant changes in BI between Day 0 and Day 12. These findings suggest that pressure-cooking is more effective in controlling browning, with no notable difference between P and CP. Chitin application in MCW requires further study, as its effectiveness appears highly dependent on the surrounding chemical environment. Overall, the combination of chitin and heat treatment showed similar effects to heat treatment alone. Despite these challenges, MCW retains potential for commercialization as a sports drink alternative due to its beneficial properties, although further research is needed to develop a product that is both appealing and nutritious
Deep learning-based classification of multichannel bio-signals for emotion recognition
No full textEmotion recognition is a critical component in advancing applications such as human-computer interaction and mental health diagnostics. While traditional methods often rely on external cues, physiological bio-signals offer a more objective measure of an individual's internal emotional state. This project presents the design, implementation, and comprehensive evaluation of a deep learning-based framework for multimodal emotion recognition, leveraging electroencephalography (EEG), galvanic skin response (GSR), electromyography (EMG), and speech audio.
The research utilized the DEAP and RAVDESS datasets to conduct a comparative analysis of different modeling approaches. Hybrid deep learning architectures, including Convolutional Neural Networks combined with Long Short-Term Memory (CNN+LSTM) and Self-Attention mechanisms, were implemented to capture spatio-temporal patterns from EEG. These were systematically compared against a benchmark model using traditional, handcrafted features (EEG Band Power, GSR/EMG statistics). To integrate information from disparate sources, both early fusion (for homogeneous physiological signals) and a novel late fusion prototype (for heterogeneous, cross-dataset signals) were developed and evaluated.
The experimental results revealed several key findings. In rigorous cross-subject validation, the traditional feature-based benchmark model demonstrated superior generalization capabilities compared to the end-to-end deep learning models, which struggled with overfitting. Concurrently, a standalone CNN model proved highly effective for classifying arousal from speech. The final late fusion prototype successfully demonstrated the ability to integrate the independently trained physiological and audio "expert" models, effectively arbitrating conflicting evidence and showcasing a viable strategy for building robust, cross-dataset multimodal systems. This project contributes a comprehensive analysis of the challenges of subject-independent classification and delivers a functional proof-of-concept for heterogeneous multimodal fusion
Real-time deep learning-based face detection and recognition with integrated liveness detection for attendance system
Get PDFThis project presents the design and development of a real-time facial recognition attendance system aimed at automating and enhancing student attendance tracking in academic settings. Leveraging advancements in Artificial Intelligence and Computer Vision, the system integrates a deep learning-based Convolutional Neural Network (CNN) that generates 1024-dimensional facial embeddings for each registered user. These embeddings are used for identity verification through cosine similarity matching, achieving reliable and high-accuracy face recognition. To address security vulnerabilities such as spoofing and proxy attendance, the system incorporates active liveness detection mechanisms, including blink detection and head movement analysis, ensuring that only live human faces are authenticated. The front-end interface enables students to register their facial data and perform attendance scanning with minimal user interaction, while the web-based backend dashboard allows lecturers to manage class sections, enroll students, and monitor attendance records. The overall system demonstrates robust performance in real-world scenarios, achieving face recognition high accuracy with consistently high precision, recall, and F1-score. SQLite is used for lightweight data storage, while the Flask framework supports the real-time backend operations. The modular architecture ensures extensibility for future improvements. While the prototype is effective for controlled environments, limitations such as dataset diversity, backend scalability, and mobile accessibility remain. Future work may focus on expanding dataset coverage, implementing a cross-platform mobile application, and upgrading to a cloud-based database for better scalability. Overall, this project serves as proof-of-concept for a secure, efficient, and deployable biometric attendance system that reduces manual effort and improves accountability in academic institutions
Smart grid: Bio-inspired algorithms energy distributions for data centers
Get PDFThe growing demand for data centre services has led to significant increases in data centre power consumption, highlighting the need for efficient power management strategies to ensure sustainable and energy-efficient operations. Virtualisation technology enables multiple virtual machines (VMs) to run on a single physical server, improving resource sharing and utilisation. However, it also introduces challenges in optimising VM placement and migration to minimise power consumption while maintaining performance. This project proposes and evaluates three bio-inspired and evolutionary algorithms for VM allocation and migration: Ant Colony Optimisation (ACO), Particle Swarm Optimisation (PSO), and a Modified Genetic Algorithm (MGA). These algorithms aim to reduce power consumption, improve resource utilisation, and enhance overall data centre efficiency. The system is implemented and simulated using the CloudSim Plus framework under both homogeneous and heterogeneous data centre environments. Four different workload scenarios were tested, and the performance of the three algorithms was compared against the data centre’s baseline VM allocation policy. Each scenario was executed 30 times to ensure the reliability and consistency of results. Simulation results demonstrate that all three proposed algorithms consistently achieved lower total power consumption across all servers compared to the baseline policy. These findings highlight the potential of bio-inspired VM allocation and migration strategies for improving energy efficiency and resource optimisation in modern data centres
Applying deep learning techniques to segmentize and classify tongue regions for traditional and complementary medicine (TCM) diagnosis
Get PDFTongue diagnosis is a fundamental component of Traditional and Complementary Medicine (TCM), yet manual inspection remains subjective and inconsistent. This study proposes a deep learning framework to enhance tongue image analysis through segmentation, classification, and explainable artificial intelligence (XAI). A Mobile U-Net model was proposed and developed for efficient and accurate tongue region segmentation. Classification tasks were conducted for both binary (stained vs. non-stained) and multi-class pathological coatings, covering clinically relevant categories. Lightweight architectures, including the proposed Efficient-ResNet, achieved competitive accuracy with minimal computational cost, demonstrating strong potential for deployment in resource-constrained environments. Grad-CAM was integrated to provide visual explanations of model decisions, improving transparency and clinical trust. Experimental results show that ResNet50 and LECA-EfficientNetV2-S achieved the highest accuracy of 99% in binary classification, while EfficientNetV2-B3 and -S excelled in multi-class tasks. Efficient-ResNet maintained strong accuracy (98.5%) with only 0.31M parameters. The findings highlight the framework’s balance of efficiency, accuracy, and interpretability, offering a practical solution to standardize and modernize tongue diagnosis in TCM for both clinical and telemedicine applications
Self-powered flexible humidity sensor based on moisture-induced electric generation
Get PDFConventional humidity sensors often suffer from limited flexibility and reliance on external power sources, hindering their integration into low-cost, sustainable, and portable systems. This study presents the development of a flexible self-powered humidity sensor based on moisture-induced electricity generation (MEG), utilizing laser-induced graphene (LIG) fabricated on lignin-coated paper. A commercially available 5 V violet laser engraver (405 nm) was employed to induce graphitization through a simple, single-step scribing process. Fabrication was optimized by varying the paper’s tilt angle (0° to 20°) to enhance voltage output and stability. Surface morphology and elemental composition were analyzed using SEM, EDX, and XRD. Under 80% relative humidity, the sample fabricated at 20° tilt angle generated the highest average voltage (~6.5 mV) but showed unstable performance. In contrast, the sample prepared at 15° produced a slightly lower voltage (~4.2 mV) with significantly greater stability, identifying it as the optimal configuration. These findings demonstrate the potential of LIG-based MEG as a cost-effective, battery-free humidity sensing solution and offer insights into scalable fabrication strategies for flexible electronics.
Keywords: humidity sensor; self-powered sensor; laser-induced graphene; moisture-induced electricity generation; flexible electronics
Subject Area: TA165 Engineering instruments, meters, etc. Industrial instrumentatio
Retirement villages in Klang Valley, Malaysia: acceptance, barriers and strategies
Get PDFMalaysia’s ageing population has increased the demand for alternative senior housing, such as Retirement Villages (RVs), which emphasise independence, security, and social engagement. While RVs have been extensively studied in
Western countries, their acceptance in Malaysia remains relatively underexplored, indicating the need for further investigation. This study aims to assess (i) the level of acceptance of RVs, (ii) key barriers to acceptance, and (iii)
strategies to enhance the acceptance. A review of existing literature revealed that RV acceptance is generally higher in Western countries compared to Asian contexts. Five key barriers were identified from literature—cultural, social,
financial, legal and technical, and the living environment—alongside four strategic areas for improvement: cultural and social adaptation, financial feasibility, legal and technical frameworks, and living environment enhancements. A quantitative research method was employed, involving the
distribution of structured questionnaires to individuals aged 30 and above who are nearing retirement or are involved in elder care decision-making. A total of 142 responses were collected and analysed using Cronbach’s Alpha reliability
test, frequency distribution, arithmetic mean, Mann-Whitney U test, KruskalWallis test, and Spearman’s correlation The findings indicated a cautious but increasing acceptance of RVs in Malaysia. While 36.6% of respondents reported
moderate familiarity with the concept, 60% viewed RVs as a viable elderly care option and would likely recommend them to others. Additionally, the study highlighted that affordability concerns were the top barrier to RV acceptance,
while enhancing accessibility and design was identified as the most effective strategy to improve acceptance. The Mann-Whitney U and Kruskal-Wallis tests revealed significant differences in acceptance levels across various social
demographics, including gender, marital status, ethnicity, education level, household income, and number of children. Spearman’s correlation test showed the strongest moderate correlation between poor environmental quality and
optimal location and accessibility. These findings provide valuable guidance for policymakers and stakeholders in improving elderly care infrastructure and promoting RVs as a viable retirement option, aligning with the Malaysia Madani
vision under the "Housing for the Rakyat" initiative.
Keywords:
retirement villages, acceptance, barriers, strategies, ageing population
Subject Area: HQ1060-1064 Aged. Gerontology (Social aspects). Retiremen
Understanding the impacts of implementing AI writing digital tools on undergraduate students’ academic writing skills
Get PDFThe use of Artificial Intelligence (AI) writing tools within higher education has become increasingly popular in terms of developing the quality of undergraduate students’ academic writing skills. This paper explores the effects of AI writing digital tools (e.g., Grammarly, ChatGPT, and Quillbot) on students’ performance in writing through a mixed-methods approach with 370 undergraduate-level surveys and semi-structured interviews. The results suggest that ease of use of the tool significantly predicts better writing skills and the need to make tools easy-to-use so as to promote engagement, while quality of feedback, writing anxiety and frequency of use are not statistically significant but rather optimistic for students. Qualitative results also confirmed that AI-tools supported immediate, non-judgmental feedback and less stress but higher confidence in writing, especially at the start of the task; however, questions were raised about generic outputs, fact-checking, lack of collaboration as well as ethical concerns around authorship. These findings have implications for writing education, as AI can help with superficial issues such as grammar, clarity, and surface fluency in writing, without reflective engagement and pedagogical mediation, the value of AI to develop deeper thinking (certainly critical thinking) and a developed argument is limited. The study indicates that AI tools require a balance with clear instructions, techniques and an ethical framework to utilize the effectiveness of AI tools in education and maintain academic integrity
Immersive interaction using augmented reality with exploratory learning approach for Lembah Bujang heritage
No full textThe promotion and preservation of cultural heritage require innovative methods that combine education with interactivity to engage and attract modern audiences. The goal of this project, Immersive Interaction using Augmented Reality with Exploratory Learning Approach for Lembah Bujang Heritage, is to create a web application that uses interactive technologies and augmented reality (AR) to improve users' educational experiences regarding Lembah Bujang's historical significance. A historical corridor that doubles as a virtual
museum, an AR module for visualizing ancient reconstructions, an interactive map for site navigation, a mini-game module to gamify learning, and a storytelling module to present captivating stories are the five key modules that make up the system. The development method places a strong emphasis on user centered design, iterative evaluation, and structured educational objectives by utilizing the ADDIE instructional design model. The project seeks to address the challenges in traditional historical site education, such as difficulty and
limited engagement in visualizing historical contexts. This can be solved by offering an exploratory and interactive learning approach. Ultimately, this application aspires to bridge cultural heritage with modern technology, fostering
greater accessibility, appreciation, and understanding of Lembah Bujang among diverse audiences, particularly young adults.
Keywords: Augmented Reality, Lembah Bujang, Exploratory Learning, Heritage Perservation, Unity, Vuforia, Blender, 3D Models.
Subject Area: QA76 – Computer Scienc
Assessing the integration of senior living technology in Malaysian real estate developments
No full textThe rapid aging of Malaysia’s population presents significant challenges to traditional housing and eldercare systems. Senior living technologies, including smart home systems, health monitoring devices, and telehealth platforms, have emerged as promising solutions to enhance the quality of life, safety, and independence of older adults. This study explores the integration of senior living
technologies in Malaysian real estate developments, focusing on their adoption, associated challenges, and impact on senior residents. Employing a qualitative and exploratory research design, semi-structured interviews were conducted with real estate developers, industry experts, and policymakers to uncover insights into the motivations, barriers, and outcomes associated with these technologies. Key findings reveal critical barriers such as financial constraints, regulatory gaps, cultural resistance, and infrastructure disparities, alongside transformative benefits in safety, health, and social connectivity. The study identifies the need for targeted strategies, including public-private partnerships, regulatory reforms, and culturally sensitive, cost-effective solutions to advance the adoption of senior living technologies in Malaysia. These insights provide a foundation for designing
inclusive, technology-enabled environments that cater to the needs of Malaysia’s aging population, contributing to sustainable societal development.
Keywords: senior living technologies; aging population; smart housing; telehealth platforms; Malaysia
Subject Area:
HT101-395 Human settlements
RA790-790.95 Geriatrics
TK7885-7895 Computer engineering
HD7287-7287.8 Housing and urban planning
HQ1060-1064 Social gerontolog