Sydney eScholarship
Not a member yet
31878 research outputs found
Sort by
Advanced Deep Learning Techniques for EEG based Sleep Signal Analysis
No full textSleep signals are crucial for diagnosing sleep disorders and analyzing sleep patterns. Polysomnography uses EEG and other physiological measurements for sleep staging and disorder detection but relies on manual analysis requiring expert knowledge, time, and cost.
While machine learning shows promise in automating sleep analysis, three gaps remain: (1) no research combines traditional signal processing theory with deep learning, (2) scarce public sleep disorder datasets limit model development, and (3) current staging focuses on macrostructure, ignoring microstructure.
This thesis addresses these gaps using single-channel EEG signals. We integrate the Hilbert transform into deep learning architecture for improved sleep staging model design. This architecture is applied to insomnia characterization, extracting features from large datasets and transferring them to smaller insomnia datasets via domain adaptation. We integrate cyclic alternating patterns (CAP), a microstructure biomarker, with sleep stages to capture finer-grained dynamics.
This thesis contributes to automated sleep staging and disorder detection by integrating signal processing theory and sleep microstructure, enabling scalable and precise tools for personalized sleep medicine
Ecology of floral visitors in Cumberland Plain Woodland undergoing restoration
No full textRestoration is key to retaining and improving the structure and function of ecosystems after disturbance. However, it is important to ensure that restoration actions result in the intended outcome. Pollination, a vital function within terrestrial ecosystems, is needed for the reproduction of flowering plants, and the long-term self-sustainability of restored areas. Insects are one group capable of performing pollination. Assessing the impact restoration has on insect floral visitors is the first step in understanding if pollination is supported. To do this, 17 sites in the Cumberland Plain Woodland (CPW), a critically endangered ecological community undergoing restoration (weeding, natural regeneration, and revegetation) were surveyed. Multiple methods were used to investigate how restoration actions are impacting floral visitors, and the possibility to support them through artificial nesting supplementation.
A seasonal one-year survey of floral visitor and flower interactions revealed that weeds and the introduced honey bee (Apis mellifera), play a prominent role in the floral visitor network. Pollard walks conducted for two years in summer, detected a higher abundance of butterflies within weeded areas than naturally regenerated areas. Deployed bee hotels, acting as artificial nesting, were primarily occupied by wasps. While bee hotels target native bees, wasps still provide important function in the CPW in the form of predation. However, the high level of parasitoid emergence, indicates that long-term deployment risks this artificial nesting supplementation becoming a population sink. These results highlight the importance of monitoring and assessment, to determine if the action taken is benefitting the functions of the restored areas. More restoration work is needed in the CPW, as the dominance by generalists and weeds indicate that this community is still impacted by disturbance
Research Paper for Australian Human Rights Commission: Experiences of Pregnancy and Pregnancy Leave in the Workplace
No full textThis report, prepared for the Australian Human Rights Commission, examines the experiences of pregnancy and parental leave in Australian workplaces since the Commission’s 2014 national inquiry. It assesses developments in legislation, workplace practice, and gender equality, identifying persistent barriers for pregnant employees and caregivers, particularly those from marginalised communities. Drawing on a human rights framework, it analyses gaps in protection and proposes reforms to expand access to paid parental leave, strengthen data collection, promote cultural change, and ensure equitable participation in the workforce. The report highlights international best practices to inform Australia’s ongoing policy and gender equality reforms
Water insecurity in Old Dhaka: A postcolonial perspective
No full textDhaka possesses sufficient freshwater to meet the needs of its entire population, but most of it flows to those in high-income neighbourhoods. In contrast, low-income communities pay significantly more for water, both financially and in terms of time and effort. This inequality is rooted in an historical legacy from the colonial period, which ingrained a bifurcated understanding of water as a commodity rather than a right. In a bifurcated system, water does not flow to those who most need it, but to those who can pay the most for it, leaving those who most need it to find their own improvised ways of accessing it.
This bifurcation has resulted in two distinct water systems: the informal system, managed by poor and middle-income people through institutions like water vendors and local Mosques, and the formal system, managed through DWASA (Dhaka Water Supply and Sewerage Authority). In line with these arguments, the research is guided by the following central questions: How has the water supply system in Dhaka failed to ensure adequate access for low-income residents of Old Dhaka? How does the current governance structure affect the political and material capacities of the urban poor? And finally, how do low-income individuals claim their right to water within the existing system? Using postcolonial and political ecology perspectives and ethnographic research in Old Dhaka, the thesis explores how government institutions bear colonial legacies that limit people's right to water and how those people deal with these limitations by asserting their right to water through a vast informal sector. Through these insights, the thesis argues that without understanding the informal water system, it is impossible to understand how low-and middle-income people assert their right to water and how they obtain water in a postcolonial city that was never configured to provide it to them
A constant state of flux: Assessing how the marketisation of community aged care in Australia influences care pricing for older Australians
No full textOver the past four decades, the marketisation of social services has reshaped the Australian welfare state. While reforms promised efficiency, consumer empowerment, and equity, little empirical evidence has tested whether these outcomes have been realised in community aged care. This thesis critically examines how marketisation has influenced care pricing for older Australians, comparing outcomes across both (i) provider types and (ii) geographic regions.
Using a comparative mixed-methods design, the study draws on secondary pricing data from the My Aged Care Provider Comparison Tool (2024) and the AIHW NextGen dataset. Two Aged Care Planning Regions in New South Wales were analysed: the Illawarra and New England. Provider-level pricing tables were constructed for all four levels of HCP funding, with benchmarks drawn from the block-funded Commonwealth Home Support Programme (CHSP).
The findings show that CDC has not reduced costs. Instead, management and administrative fees consume a significant proportion of HCP funding, resulting in higher care costs for consumers. Geographic disparities are pronounced: competitive conditions in Illawarra drive lower costs per hour (around AUD 100 per hour. Provider type also matters, noting that for-profit organisations generally deliver more hours of care per dollar than charitable, community-based, or government providers. These outcomes highlight that marketisation produces uneven benefits, raising concerns about efficiency and equity.
The thesis concludes that stronger market stewardship is essential. Policy responses should include standardised fee limits, adjustments for rural cost differentials, and improved transparency to ensure fair access. These lessons are timely given the pending unification of CHSP and HCP into the Support at Home Program, which risks replicating existing inequities unless reforms are carefully designed
The experiences of caregivers of individuals with head and neck cancer: a qualitative study
No full textStudy background
Head and neck cancer (HNC) is a devastating disease. Approximately 20 million people were diagnosed with cancer worldwide in 2022, which is expected to become 35 million by 2050. HNC treatments are often complex, and their associated side effects can result in substantial functional challenges and visible changes for patients. Depending on individual circumstances, recovery from the treatment side effects can last weeks to months or even longer. These patients require considerable care, and caregiving tasks can be challenging for the caregivers. To date, little is known about the experiences of the informal caregivers of individuals with HNC around the world, including Australia, to identify their support needs.
Methods
This study was guided by qualitative interpretive inquiry. The research setting was a metropolitan tertiary hospital in New South Wales (NSW), Australia. Fifteen participants were recruited for this study from 2020 to 2022. Semi-structured individual interviews were conducted via telephone with 15 caregivers. The interviews were analysed using a thematic analysis.
Findings
Becoming a caregiver for someone with HNC was a life-changing responsibility, and this transition was emotionally and psychosocially challenging, making caregiving one of the most difficult roles participants had ever assumed. The participants experienced changes in their daily lives as they adjusted to new roles and responsibilities. Those changes included psychosocial impacts, relationship changes and disruptions to their daily routines. The participants acted as coaches in providing care for HNC patients. The caregiving role involved providing emotional, social and financial support; facilitating communication between patients and others; and monitoring patients’ symptoms and overall well-being. Many participants in this study gained a sense of purpose in managing familiar and unfamiliar care needs when caring for the person with HNC
Advanced Topology, Modulation, and Control for Achieving High-Performance Solar Microinverters
No full textSolar microinverter systems are classified into two-stage and single-stage topologies. Two-stage designs are constrained by the poor performance of the second-stage full bridge inverter (FBI), while single-stage dual active bridge (DAB) microinverters suffer from the lack of active power decoupling (APD).
Chapter 2 proposes a DAB+FBI topology with an active auxiliary branch, achieving zero-voltage turn-on switching, fixed frequency, and continuous conduction mode (CCM) FBI operation. Simulations show a 60% loss reduction and 98.14% peak efficiency.
Chapter 3 introduces an APD modulation that reuses the primary-side DC-blocking capacitor as a decoupling capacitor with a large-signal control loop, cutting capacitor volume by 75% and input current ripple by 96%.
Chapter 4 develops a synergistic modulation extending APD to non-unity power factor. With an inner phase shift, secondary current is optimized, reducing RMS current by 27% and improving efficiency by 4.5%, reaching 93.8% full-load efficiency.
Overall, the proposed topologies and modulation strategies enhance efficiency, power density, and MPPT accuracy, advancing high-performance solar microinverter systems
AI Assess AI: Automating Perceptual Quality Assessment for AI-Processed Immersive Media
No full textAs AI technologies proliferate, AI-processed content (e.g., frame interpolation, ChatGPT) is everywhere. Yet judging its quality is critical. Subjective experiments, though time-consuming, yield limited, dataset-specific opinions, highlighting the need for automated QA with minimal labels.
Immersive media—VR, AR, MR—uses Light Field Imaging (LFI) and 3D reconstructions to create realistic, interactive experiences. Although methods like PSNR, SSIM, and LPIPS exist for 2D QA, immersive media remains underexplored. This thesis advances perceptual QA for immersive media, from LFI to 3D reconstructions.
Automatic QA is either full-reference or no-reference, but most modern content lacks references, making no-reference QA more relevant. First, I develop no-reference LFIQA. Because LFIs add an angular dimension, I adapt depthwise separable convolution for spatial features and propose anglewise separable convolution for comprehensive quality assessment. I also introduce anglewise attention—multihead self-attention, grid attention, and central attention—to improve feature extraction at lower complexity.
Next, I propose the first no-reference QA for photorealistic 3D reconstructed scenes from NeRF-like or 3D Gaussian methods. My approach integrates viewwise (spatial) and pointwise (angular) assessments, evaluating each synthesized view’s quality and inter-view consistency, plus the angular properties of surface points. Experiments confirm these no-reference methods outperform existing approaches, laying the groundwork for robust perceptual QA in immersive media
Direct strength method for the design of cold-formed steel hat sections under localised loading
No full textThis thesis investigates the application of the Direct Strength Method (DSM) in predicting the localised loading capacity of cold-formed steel hat sections. Four loading cases with various fastening conditions are studied. Experimental data from this study and previous research are utilised to develop a yield/plastic mechanism model for use in the DSM, and to propose simplified and consistent DSM design equations for practical applications. The potential for utilising the DSM design equations for localised loading design is established through numerical, experimental and analytical investigations. Methods for calculating the two significant inputs for DSM design equations, the buckling load (P_cr) and the yield load (P_y), have been investigated.
The Finite Element Method (FEM) and the Finite Strip Method (FSM) are widely accepted for calculating the buckling loads of channel sections. A software program called THIN-WALL-PYTHON is developed and optimised for hat sections, employing the FSM to calculate the buckling loads which are benchmarked with the results obtained using the FEM software (ABAQUS).
Simplified yield/plastic mechanism models are developed for use with the hat sections under localised loading in various cases to calculate the yield load for DSM design equations. Experiments on hat sections with varying dimensions, particularly in the stocky range, are conducted to verify the analytical theory and numerical models.
Two sets of DSM design equations for hat sections under localised loading are developed based on experimental and numerical datasets. Deviation and reliability analyses are conducted using the experimental data, and design capacity reduction factors for both methods are proposed for all the investigated loading cases. The prediction results are compared with current standards and specifications (AS4600:2018 and AISI S100-16), demonstrating significant advancements in accuracy and applicability