48080 research outputs found
Sort by
HEArgmax: Secure homomorphic encryption-based protocols for Argmax function
No full textIn the current era of big data, cloud-based Machine Learning as a Service (MLaaS) – where clients send encrypted queries to the cloud and receive prediction results – has gained significant attention. However, privacy concerns arise as cloud servers typically require access to clients’ raw data, potentially exposing sensitive information. Homomorphic encryption (HE), an advanced cryptographic technique that allows computation on encrypted data without decryption, offers a promising foundation for privacy-preserving MLaaS. A critical challenge in this context is the efficient and secure evaluation of the argmax function—a key operation in classification tasks used to select the class with the highest predicted probability. Existing HE-based methods, such as Phoenix (Jovanovic et al., 2022), rely on non-interactive protocols using high-degree polynomial approximations of the sign function, which lead to significant computational overhead. This paper introduces HEArgmax, an interactive protocol designed for efficient and secure argmax evaluation under encryption. Unlike prior approaches, HEArgmax leverages the algebraic properties of the sign function in combination with a lightweight interactive mechanism under the standard semi-honest model, without requiring trusted setup or multi-party computation. We present two protocol variants: HEArgmax-HT, optimized for high-throughput scenarios using batch processing, and HEArgmax-LC, which minimizes communication by processing a single encrypted vector. Experiments show that HEArgmax reduces inference latency from 157 s to 8 s on the MNIST dataset, and performs well even on CIFAR-100 with 100 output classes, completing in under 4 min using 128-bit HE security parameters. Despite being interactive, our protocol achieves comparable communication costs to Phoenix. These results demonstrate that HEArgmax is both practical and scalable for real-world privacy-preserving MLaaS deployments.</p
Rethinking microentrepreneurship: Bricolage and resilience among Goroers in Accra's informal economy
No full textEntrepreneurship is widely recognised as a pathway to alleviating global inequality and poverty. Yet mainstream literature often overlooks the survival-driven entrepreneurial practices of people in poor and developing regions, particularly sub-Saharan Africa. Microbusinesses in these contexts are frequently excluded from general entrepreneurship surveys or relegated to an ‘other’ category, limiting understanding of their unique contributions.This study contributes to the entrepreneurship-as-practice movement by examining Goroers’ everyday activities through the lens of entrepreneurial bricolage. It challenges the conventional focus on high-growth firms, showing how microentrepreneurship in Accra is shaped by social values, political realities, and cultural identity. Using a processual ontology, performative epistemology, and case-study method, the research explores how resource constraints continually reshape entrepreneurial processes and how bricolage enables entrepreneurs at society’s margins to navigate these challenges.Data was collected from 20 microentrepreneurs (Goroers) and 15 key informants in Accra through in-depth interviews, observations, and reflective practices. Thematic analysis revealed that despite structural challenges, such as unemployment, income inequality, and inadequate infrastructure, entrepreneurship persists. These obstacles often propel the emergence of Goroing as a series of adaptive strategies in a context of resource constraints. The study shows how Goroers mobilise resources through spatio-temporal bricolaging, enforceable trust, bounded solidarity, mentorship, diversification, and continuous learning. Beyond structural challenges, Goroers also face hostile discourses and systemic neglect, including negative portrayals, infrastructural deficiencies, and recurrent evictions, which destabilise their work and shape both public image and self-identity. Yet, Goroing fosters individual empowerment, economic inclusion, and community resilience. From these insights, the study develops a heuristic for understanding microentrepreneuring in sub-Saharan Africa, demonstrating that such practices not only support survival but also contribute to broader economic and social stability in Accra.</p
On Automated Testing of Web Applications
No full textWeb applications have become omnipresent in modern society, and their quality assurance is critical. However, automating testing of such applications — particularly due to the visual and subjective nature thereof — remains challenging. Two persistent problems hamper this: the oracle problem and the test fragility in regression testing.In this thesis, we propose novel approaches to address these automation challenges in the context of modern CI/CD workflows, leveraging failure patterns, domain knowledge and neural networks.Motivated by academic–industry collaboration and real-world needs, two core approaches are adopted. The first adapts Adaptive Random Testing (ART) to the output space, using failure-pattern heuristics to improve the effectiveness and efficiency of layout fault detection. The second enhances automated visual oracles using neural networks trained on synthetic UI mutations guided via Metamorphic Testing (MT) principles.The findings show that domain-knowledge failure-based testing can significantly outperform conventional strategies in both speed and effectiveness in detecting layout regressions. Meanwhile, the AI-enhanced oracle shows strong semantic comparison, achieving high precision and effectiveness in detecting meaningful visual defects while minimizing false positives. Designed to be lightweight and practical, both approaches are compatible with CI/CD workflows using standard hardware.While acknowledging that testing automation remains an ongoing challenge, this research presents a realistic and optimistic pathway forward — demonstrating a combination of techniques, domain knowledge test design and practical implementation can make automation more resilient, scalable, and aligned with quality expectations.</p
Towards a trustworthy internet of vehicles: Security-driven decentralized federated learning frameworks for vehicular networks
No full textThe convergence of vehicular technology, artificial intelligence (AI), and distributed computing has catalyzed the emergence of the Internet of Vehicles (IoV) as a cornerstone of next-generation intelligent transportation systems (ITS). By enabling vehicle-to-everything (V2X) communication, IoV supports cooperative perception, real-time decision-making, and autonomous driving. However, the reliance on large-scale, data-driven intelligence in IoV exposes systems to critical challenges, including adversarial poisoning, privacy leakage, identity forgery, and the fragility of centralized learning architectures. Federated Learning (FL) has been proposed as a promising paradigm to alleviate some of these issues by enabling distributed model training without centralizing sensitive vehicular data. Nonetheless, conventional FL remains vulnerable to security and trust limitations, particularly in dynamic vehicular environments. This thesis addresses these challenges by designing secure, privacy-preserving, and scalable FL frameworks that leverage distributed ledger technologies and cutting-edge security mechanisms.The thesis advances knowledge through four interconnected contributions. First, two novel optimization-driven poisoning attack models are introduced: PA-PSOSA and PAPSOGA, which combine particle swarm optimization with simulated annealing and genetic algorithms, respectively. These models demonstrate that even a small poisoning budget can substantially degrade global model utility under black-box and clean-label constraints, highlighting the urgency of robust defenses in vehicular FL. Second, a permissioned blockchain-enabled FL (BCFL) framework is proposed, in which consortium edge nodes running Practical Byzantine Fault Tolerance (PBFT) consensus replace the central aggregator. With blockchain integration and data validation mechanisms, this design ensures identity authentication, verifiable audit trails, and improved resilience against poisoning and Sybil attacks, while maintaining high model accuracy under adversarial conditions. Third, the framework is further enhanced to achieve inference-resistance by integrating secure aggregation (SecAgg) and differential privacy (DP), and lightweight with off-chain commitments. This design significantly reduces ledger storage requirements, increases system throughput, and mitigates inference-based privacy risks. Finally, to overcome the scalability limitations of PBFT-based BCFL, a DAG-enabled FL (DFL) framework is developed. By leveraging parallel validation, utility-score-based tip selection, and reputation-weighted aggregation, this framework significantly improves scalability, reduces communication complexity, and enhances robustness in asynchronous vehicular environments.Together, these contributions articulate a coherent progression from exposing vulnerabilities in vehicular FL to constructing secure, privacy-preserving, and scalable frameworks tailored for IoV ecosystems. The findings demonstrate that interdisciplinary integration of optimization theory, cryptography, differential privacy, and distributed ledger technologies is indispensable for trustworthy vehicular intelligence. Beyond theoretical significance, the proposed frameworks offer practical designs for deployment in safety-critical IoV environments. Future research directions include the integration of zero-knowledge proofs (ZKP) for verifiable privacy, adaptive defenses against evolving adversarial strategies, and experimental validation in real-world vehicular testbeds. Collectively, this thesis establishes a foundation for secure federated intelligence in IoV, contributing to the reliability, efficiency, and trustworthiness of next-generation ITS.</p
The Adoption of Electronic Arbitration in Saudi Arabia: Analysing Challenges and Catalysts in the Context of COVID-19 – A Case Study of the SCCA and Comparison with ACICA
No full textCommercial disputes are often resolved through arbitration, which offers efficiency and flexibility compared to court litigation. The advancement of digital technologies and the rise of electronic transactions have led to the emergence of electronic arbitration as a modern dispute-resolution method. This thesis examines the adoption of e-arbitration in Saudi Arabia, assessing its challenges and catalysts within the broader context of Saudi Vision 2030 and global trends in digital dispute resolution.This research addresses a critical gap in legal and academic literature by determining whether Saudi Arabia’s current arbitration framework can support the transition to fully electronic arbitration. While traditional arbitration functions effectively within the Kingdom, electronic arbitration remains underdeveloped, particularly in terms of legal recognition, procedural efficiency, and enforceability of arbitral awards. This study argues that, despite existing structural and cultural challenges including limited codification of Sharia principles, reliance on traditional mediation practices, and gaps in digital infrastructure, Saudi Arabia is making substantial progress toward adopting electronic arbitration. Catalysts such as the flexibility inherent in Sharia law, the legal reforms introduced under Vision 2030, and the pressures of the COVID-19 pandemic have accelerated this transition.Hence, for this study, a comparative legal analysis and a case-study approach is adopted focusing on the Saudi Centre for Commercial Arbitration (SCCA) and its electronic arbitration framework in comparison with the Australian Centre for International Commercial Arbitration (ACICA). It also reviews Saudi arbitration laws, international arbitration standards, and legal reforms, with particular attention given to the impact of the Arbitration Law 2012 (Saudi Arabia) on the development of electronic arbitration.Findings indicate that despite significant digital transformation efforts under Vision 2030, major obstacles persist, including insufficient codification of electronic arbitration procedures, the enforcement of electronic arbitral awards, and gaps in digital infrastructure. However, three key catalysts — COVID-19, regulatory advancements, and global best practices are driving reform efforts.The thesis proposes several recommendations for strengthening electronic arbitration in Saudi Arabia, including clearer legislative provisions, enhanced judicial training, and investment in secure digital platforms to improve efficiency. By drawing from the Australian model, the research offers reform suggestions that balance Sharia compliance with international arbitration standards.</p
An Investigation of the Nature and Extent of Human‐Robot Collaboration in Australian Industries
No full textHuman-Robot Collaboration (HRC) is an emerging field of research. Human-Robot Collaborations pose challenges for Human-Robot Interaction from a Human Factors (HF) perspective, many of which are only now being identified. Robot hardware capabilities and the sophistication of the algorithm protocols can impact the safety and efficiency of the robot interacting with the human.A narrative literature review identified limited evidence regarding the human capabilities, knowledge, and skills within human-robot collaboration both internationally and in Australia.This four (4) stage Master of Philosophy project investigated this relationship concerning the Human Factors elements of Human-Robot Collaboration.Stage 1 (Study 1) investigated the nature and extent of use of collaborative robots (cobots) in industry in Australia, using a mixed methods approach with an anonymous online survey (administered via Qualtrics) (n=5) and virtual face to face interviews with robotics academics, experts/ researchers, manufacturers & industry representatives (n=5). The study findings included: cobot users were principally large multinational companies; tasks mainly ‘pick and place’ and welding applications; main industries where cobots are used were food and packaging industries, and fabrication sector.Stage 2 (Study 2) investigated operators experiences with human-robot collaboration in industries in Australia. An exploratory mixed methods approach involved two parts: (Part A) a survey (n=3) and (Part B) a Focus Group discussion (n=3) with cobot operators. The study identified advantages of cobots included performance of tasks inherently difficult, monotonous or repetitive. Efficiencies of cobots included precision, reliability and repeatability. Stage 3 (Study 3) evaluated operators’ performance with a collaborative robot (cobot). This study utilised an observational methodology with experienced (n=3) and novice (n=8) operators. Operators performed ‘pick and place’ tasks with a cobot. Findings suggested experienced operators were more inclined to problem solve faults using a top-down problem-solving approach primarily interacting with the control tablet software program interface, whilst novices used operational bottom-up problem-solving methods such as trouble shooting using the ‘Stop’ ‘Reset Start’ control buttons to rectify a forced stop. Both experienced and novice operators were efficient in their problem-solving strategies.Informed by the findings of Stages 1, 2 and 3, Stage 4 provided recommendations for improving training strategies for operators who work with collaborative robots (cobots) in industry that could potentially increase the efficiency of the human contribution in collaborative tasks performed with a cobot.</p
Development of Self-Compacting Ultra-High-Performance Geopolymer Concrete
No full textThe abstract for this item has not been populated.</p
Advancing Deep Learning Methods for Long-tailed Image Recognition
No full textOne of the most important factors for the success of any deep learning model is the distribution of data. The effectiveness of deep learning models is significantly affected by the distribution of data within the training data set. Deep learning algorithms have already achieved outstanding performance in terms of accuracy and efficiency on the balanced distribution of data; however, if the data distribution in the dataset is imbalanced, i.e., long-tailed, it results in a biased classifier causing the degraded performance of the deep learning model. Long-tail distribution-based image recognition has attracted much attention from researchers in the field of computer vision since the last decade. Different methods have been proposed to address the long-tail distribution-based image recognition problem in the existing literature, however, there is still room of improvement. So, this research is aimed to address the problem of long-tail distribution-based image classification for deep learning models. In particular, this research unfolds its investigation into how to mitigate the long-tailed data distribution challenges against the biased classification from the following three aspects.Firstly, this thesis argues that the diversity of image features, inter-class distance and intra-class distance play a vital role in image recognition and enables the classifier to classify the images correctly. Accurate classification requires well-separated classes (large inter-class distance) and compact sample distributions within each class (small intra-class distance). In long-tailed settings, tail classes contain few and less diverse samples, which leads to larger intra-class distance and consequently reduces their discriminability. So, this thesis has proposed and implemented a novel method of enriching the feature space of tail-class samples, i.e., incorporating the variance of angles from the head-classes to the tail-classes with supervised contrastive learning. Performing experiments on different long-tailed benchmark datasets shows that this method has also outperformed many of the existing baseline methods.Secondly, this thesis introduces a novel approach that augments the traditional feature scaling mechanism by incorporating adaptive margin constraints into the scaling factor itself, thereby enabling dynamic expansion of the feature space throughout training. In conventional long-tailed recognition, fixed or uniformly applied scaling often fails to sufficiently separate minority class features from those of the majority, especially in high-dimensional embedding spaces where class boundaries tend to collapse under imbalance. This thesis addresses this by embedding class-dependent adaptive margins within the scaling formulation, allowing the model to not only compact intra-class features but also actively push inter-class boundaries outward in a controlled manner. This adaptive margin is updated jointly with the network parameters in an end-to-end optimization process. Empirical evaluations demonstrate that this adaptive margin integration consistently boosts Top-1 accuracy, F1-score, and AUC across multiple long-tailed benchmarks, confirming its effectiveness to existing classification strategies.Finally, this thesis addresses the joint presence of long-tailed distributions and noisy labels. The proposed framework, i.e., Hybrid Distribution-aware Graph and Prototypical Contrastive Learning (HDG–PCL) departs from purely batch-wise remedies by maintaining a global memory bank that stores high-confidence samples identified via the consensus of a conventional classifier and a balanced classifier throughout training. These global prototype and local graph (hybrid) objectives are optimised alongside supervised and mixup-based semi-supervised losses, culminating in balanced, noise-tolerant representations. Extensive experiments on long-tailed datasets demonstrate that HDG–PCL sets a new state of the art under simultaneous presence of imbalance and noise factors.In summary, this thesis investigates, designs, and implements advanced methodologies to mitigate the adverse effects of long-tailed data distributions on deep learning performance. Through the introduction of three novel contributions, i.e., Supervised Angular Contrastive Learning with Balancing Classifier (SACL-BC), Optimized Adaptive Feature Compression (OAFC) to enhance feature space geometry, and Hybrid Distribution-aware Graph and Prototypical Contrastive Learning (HDG–PCL) that synergistically combines global and local representations, this work advances the state of the art in achieving higher image classification accuracy under severe class imbalance conditions. Despite the demonstrated effectiveness of the three proposed contributions, each method carries certain limitations. First, the feature-space enrichment strategy (SACL-BC) requires the use of a manually selected threshold to separate head and tail classes for different imbalance ratios, which may limit its adaptability across datasets with varying distributions. Second, the adaptive-margin scaling approach (OAFC) introduces additional hyperparameters and relies on stable margin updates during training; hence, its performance may be sensitive to margin initialization and optimization settings, potentially requiring careful tuning for different architectures or datasets. Third, although the HDG–PCL framework effectively handles the joint presence of imbalance and noisy labels, maintaining a global memory bank increases computational and memory overhead, and the method’s reliance on consensus-based pseudo-labeling may still be affected when both classifiers exhibit high uncertainty in extremely noisy scenarios.</p
Supraharmonic Emissions in Future Electricity Networks with Solar PV Systems
No full textThe world is experiencing a shift from centralised to decentralised power generation as a pivotal strategy for achieving net-zero emission targets. As a result, solar photovoltaic (PV) systems have become the leading distributed energy resource (DER) connected to the grid. The integration of new DER systems presents several technical challenges, one of which is the power quality (PQ) issues introduced by associated power electronic interfaces. Conventional PQ studies have concentrated on voltage and current waveform distortion in the frequency range of 0–2 kHz, however, special attention has been paid to conducted emissions above 2 kHz over the past decade since their systematic knowledge is not completely developed yet. Waveform distortion in the 2–150 kHz frequency range is referred to as “supraharmonic emissions” or “high-frequency harmonic emissions”.This research study aims to identify the supraharmonic emission characteristics from rooftop solar PV systems containing identical PV inverters and their propagation characteristics within low voltage (LV) networks, focusing on different supraharmonic phenomenon appearing within the 2-25 kHz frequency range.The first part of this study is aimed at characterising supraharmonic emissions based on measurements of an LV installation with a rooftop solar PV system, consists of six (6) identical inverters. There are three measurement methods stipulated in international regulations, i.e. IEC 61000-4-7 (informative), IEC 61000-4-30 (informative), and CISPR 16-1-1 (normative). However, those regulations do not include prescriptive measurement and analysis techniques for the entirety of this range (2-150 kHz). Thus, the measurement methodology in IEC 61000-4-7 which is intended for emissions in the 2-9 kHz range, is utilised to analyse measurements up to 25 kHz in this study maintaining the consistency between 2-9 kHz and 9-25 kHz ranges.Two supraharmonic emission phenomena have been identified in two distinct frequency ranges: (a) resonance related emissions (2-5 kHz) due to the front-end filter of PV inverters, and (b) switching frequency emissions (15-17 kHz) due to the high frequency switching PWM pulses. Simultaneous measurements at different locations of the same LV installation showed that the emissions associated with resonance propagates within the installation, however, they do not propagate into the substation level where electricity is supplied to the installation. In addition, three (3) methods have been proposed to represent the phase angle for grouped supraharmonic emissions to be incorporated with harmonic summation techniques, which will assist in understanding the phase angle diversity of supraharmonic emissions.The second part of the study is aimed at analysing supraharmonic emissions based on a generic MATLAB/Simulink model. The model is developed using the commonly adopted voltage oriented control (VOC) under synchronous reference frame, often referred to as direct-quadrature (d-q) reference frame, identifying the boundary up to which low-frequency harmonic models are capable for supraharmonic analysis. The model is capable of representing the resonance and switching frequency supraharmonic emission phenomena identified, and is validated based on the measurements used in the first part of the study. The impact of carrier waveform phase shift in PWM signal generation is investigated within a multiple solar PV inverter system, which can be used to mitigate switching frequency harmonic flow into the grid.The study uniquely delves into characterising these emissions using advanced methods, such as phase angle representation, and improving understanding of the complex interactions that occur at the micro-level of inverters. Thus, the results from this research study are of significant importance to characterising and developing systematic knowledge on supraharmonic emissions from rooftop solar PV inverter systems.</p
Experiences and support needs of families affected by a loved one’s alcohol or other drug use
No full textThe abstract for this item has not been populated.</p