Work-in-Progress Paper - NPS Australia Submission System

Development and Preliminary Evaluation of Banana Stem Fiber-Based Biocomposite Using Natural Okra Mucilage and Rice Gum Binders

This study develops a novel bio-composite from banana stem fiber and coconut particulate filler by the use of two food grade natural binders such as okra mucilage and rice gum paste as fully plant based alternatives to synthetic resins. Through direct qualitative comparison, okra mucilage was found superior to rice gum paste in rigidity, surface quality and fiber binder adhesion. It’s also retaining strong adhesion under water and oil exposure. The work establishes a locally sourced low cost and biodegradable composite system aligned with circular economy principles and UN SDGs 9, 12, and 13 laying the foundation. This is a continuous process, the strength of the newly formed composite is yet to confirm, however the composite shows better adhesion.

Designing a Prompt Framework for Generative AI–Based Critical Thinking Support in SSC Mathematics Learning

As Generative Artificial Intelligence (Generative AI) gains traction, it is poised to offer
innovative solutions for teaching and learning mathematics, particularly in the form of intelligent tutoring systems. But most current AI tutoring systems are designed to output
answers rather than to help students through a sequence of reasoning and problem-solving
steps. For this reason, their contribution to the development of critical thinking is limited,
a very important goal of mathematics education. In addition, current prompt engineering
techniques tend to be optimized for a single prompt rather than offering a complete pedagogical structure based on validated educational theory.
This study aims to develop a structured prompt framework for a Generative AI tutoring
system for facilitating critical thinking in the learning of mathematics in Secondary School
Certificate (SSC). The framework combines the problem solving approach of Pólya with
prompts representing the four stages of the problem solving process: understand the problem, make a plan, execute the plan, look back. Prompts for learners’ guidance, evaluation of
their responses, adaptive feedback, and advancing to the next stage are included as distinct
sections in each stage to facilitate the learners’ active reason and reflective learning.
The study is guided by Design and Development Research (DDR) methodology consisting of four phases namely problem identification by literature review, design of framework,
design of prototype and evaluation by expert opinion. A prototype of AI mathematics tutor based on the proposed framework is built, and the framework is assessed by experts in
computer science and mathematics education through questionnaires and semi-structured
interviews. The assessment is on the clarity, pedagogical relevance, usability and potential
of the framework to support critical thinking.
This inquiry offers a pedagogically informed prompt model incorporating the principles of mathematics learning and of prompt engineering for Generative AI. The proposed
framework offers concrete design considerations for designing AI tutoring systems to support structured mathematical thinking and critical thinking, especially in the context of SSC
mathematics learning.

An Explainable AI-Powered Performance Weighted Ensemble Deep Learning System for Classifying MRI Brain Tumours

This work investigates the classification of brain tumors from MRI scans using a variety of deep learning architectures, such as Light U-Net, DenseNet121, Attention U-Net, and Custom Convolutional Neural Network (CNN). To increase accuracy, an ensemble-based classifier was created with a weighted soft voting technique. Explainable AI (XAI) approaches, like Gradient-weighted Class Activation Mapping (Grad-CAM), were used to improve clinical applicability by visualization. With an accuracy rate of 97.93\% in differentiating between gliomas, meningiomas, pituitary adenomas, and healthy patients, the ensemble model behaved better than single structures. Heatmaps produced by Grad-CAM showed important areas affecting model predictions, matching radiological characteristics and boosting confidence in the outcomes.

ComML: A Code Comments Dataset for AI/ML Systems

Code comments play a critical role in ensuring
the comprehensibility, maintainability, and collaborative devel-
opment of software systems. However, in Artificial Intelligence
and Machine Learning (AI/ML) projects, where experimentation,
rapid prototyping, and iterative model development are cen-
tral, commenting practices are often inconsistent or insufficient.
To support systematic research on this issue, we introduce
ComML, a real-world dataset for analyzing code comments in
AI/ML systems. ComML comprises 50 AI/ML repositories written
in Python, having 16,110 Python files, 3,314 classes, 111,950
functions, 146,264 code blocks, 5,662,809 lines of source code,
and 449,962 lines of comments. The dataset was constructed
by cloning repositories from GitHub and extracting structured
code blocks using Python’s Abstract Syntax Tree parser. Each
code block was automatically classified into one of nine AI/ML
workflow stages. Then we checked the frequency and quality
of the comments inside the code. Along with raw code and
comments, ComML provides computed metrics for comment
density, readability, and consistency. Our initial analysis reveals
notable documentation gaps, with the Model Evaluation Stage
missing comments in 74.09% of cases. This dataset enables
researchers to study commenting behavior across AI/ML de-
velopment phases and supports the design of automated tools
for improving comment quality and documentation practices in
machine learning software.

Bone Fracture Detection And Localisation In X-Ray Using Real Time Object Detection Model

The proposed
model achieved a mean Average Precision at IoU 0.50 (mAP50) of 0.93, which represents a significant improvement over established benchmarks —a 9.4% relative increase over YOLOv5 (0.85 mAP50) and a 13.4% relative increase over Faster R-CNN (0.82 mAP50). It also demonstrated high precision (0.91) and recall (0.92), indicating robust performance in accurately identifying and localising fractures with a low rate of false positives and negatives.

Bone Fracture Detection And Localisation In X-Ray Using Real Time Object Detection Model

The proposed
model achieved a mean Average Precision at IoU 0.50 (mAP50) of 0.93, which represents a significant improvement over established benchmarks —a 9.4% relative increase over YOLOv5 (0.85 mAP50) and a 13.4% relative increase over Faster R-CNN (0.82 mAP50). It also demonstrated high precision (0.91) and recall (0.92), indicating robust performance in accurately identifying and localising fractures with a low rate of false positives and negatives.

A Transformer-Based Multimodal Framework for Enhanced Autism Spectrum Disorder Diagnosis

1) To improve identification performance, we proposed a multimodal framework that integrates medical imaging and clinical textual data.
2) Introduction of a classification token mechanism to enhance feature representation and determination with Vision Transformer.
3) Finally, we implement hyperparameter optimization techniques to improve model efficiency, generalization, and overall performance.

Solar-Powered IoT-Based Smart Farming Model (SPISFM)

Title: Leveraging Web Applications for Enhanced Transportation Mobility: Integrating Taxi Booking and Volunteer Ride Services in Fiji

The significant research contribution of this project is the development of a web-based platform that integrates real-time taxi booking, ride-sharing, and volunteer ride services tailored for Fiji. This innovative solution addresses key challenges in Fiji’s urban transportation, such as traffic congestion, vehicle overuse, and lack of affordable transport for low-income individuals. By promoting environmental sustainability, fostering community engagement through volunteer rides, and leveraging secure online payment systems, this platform contributes to enhancing mobility and reducing greenhouse gas emissions in a unique socio-economic context.

Leveraging ChatGPT for Sponsored Ad Detection and Keyword Extraction in YouTube Videos

This study is significant for several reasons. First, it provides a scalable and automated solution for detecting and analyzing advertisements within video content, which is typically labor-intensive when done manually. Second, it offers insights into the relationship between advertisements and video content, which can have profound implications for advertisers seeking to improve targeting strategies and for content creators aiming to optimize sponsored ad placements within their videos. Third, the research lays the groundwork for future advancements in content-based advertising, where the alignment between ad messaging and content themes can be refined using advanced natural language processing (NLP).

Leveraging Web Applications for Enhanced Transportation Mobility: Integrating Taxi Booking and Volunteer Ride Services in Fiji’s

The significant research contribution of this project is the development of a web-based platform that integrates real-time taxi booking, ride-sharing, and volunteer ride services tailored for Fiji. This innovative solution addresses key challenges in Fiji’s urban transportation, such as traffic congestion, vehicle overuse, and lack of affordable transport for low-income individuals. By promoting environmental sustainability, fostering community engagement through volunteer rides, and leveraging secure online payment systems, this platform contributes to enhancing mobility and reducing greenhouse gas emissions in a unique socio-economic context.

A Study on Object Detection Performance through Data Augmentation under Adverse Weather Conditions

This study compares the performance of object detection models through data augmentation with a severe weather dataset.

PREDICTING THE CUSTOMER BEHAVIOR UTILIZING TREE BASED MACHINE LEARNING ALGORITHMS

The goal of this project is to predict customer behavior from a large real-world e-commerce dataset using tree-based machine learning modeling techniques that will employ decision tree, random forest, and gradient boosting. Each of the models will be evaluated and compared to determine which of the three is the best model for predicting customer behavior.

PREDICTING THE CUSTOMER BEHAVIOR UTILIZING TREE BASED MACHINE LEARNING ALGORITHMS

The goal of this project is to predict customer behavior from a large real-world e-commerce dataset using tree-based machine learning modeling techniques that will employ decision tree, random forest, and gradient boosting. Each of the models will be evaluated and compared to determine which of the three is the best model for predicting customer behavior.

Review of Mathematical Modelling and Interference Minimization Schemes for the Coexistence of 5G and Satellite Radio Access Networks

The aim of the study is to develop a suitable algorithm for interference minimizing in 5G and satellite communication networks coexistence employing Nakagami-m and Shadowed Rician models. Based on this aim, the following research contributes to:
1-Develop a suitable theoretical strategy that evaluates interference scenarios for co-existence between5G and satellite communication networks.
2-Develop an algorithm based on Nakagami-m and Shadowed Rician models for interference minimization in the co-existence between 5G and satellite communication networks.

Patterns In Twitter Use During a Disaster: Content Analysis of 2023 Türkiye-Syria Earthquake Tweets

We analyze more than 400,000 tweets posted between 6-21 February 2023, and explore different use cases of Twitter networking site aftermath of the quake series. We carry out descriptive analysis of the tweets distribution, and analysis on hashtag agenda setting property. Topic distribution both in hashtags and tweet content is investigated.

Evaluating Lightweight Asymmetric Cryptography for Secure Communication in Internet of Drones

Unmanned aerial vehicles (UAVs) are being successfully used in a variety of applications, including agriculture, search and rescue operations, surveillance systems, and mission-critical services, thanks to some technological and practical advantages, such as high mobility, the ability to extend wireless coverage areas, or the capacity to reach locations inaccessible to humans. In contrast, attacks against drones, as opposed to traditional cyberattacks, typically happen as a result of serious design flaws and a lack of wireless security protection methods. The study examines lightweight asymmetric cryptographic algorithms for secure Internet of Drones (IoD) communication, addressing cybersecurity
challenges within this emerging technology. It evaluates RSA, ElGamal, DiffieHellman, and Elliptic Curve Cryptography (ECC), focusing on their suitability for IoD through comparative analysis on calculation time, memory usage, key size, and security. The goal is to contribute to developing robust, efficient, and secure communication protocols for IoD, promoting growth while mitigating risks. This research is pivotal for the advancement of IoD security, exploring the application of these cryptographic techniques to ensure secure, efficient operations within the IoD framework.

Paper Type: Work-in-Progress Paper