Sydney eScholarship
Not a member yet
31878 research outputs found
Sort by
Chineseness in the making: An ethnographic linguistic landscape study of Chinese restaurants in Hurstville, a Sinoburb of Sydney
No full textThis thesis examines how Chineseness is produced, negotiated and projected through the linguistic landscape of Chinese restaurants in Hurstville, a major Chinese-concentrated suburb, or Sinoburb, in southern Sydney. It addresses a gap in sociolinguistic research by treating ethnic restaurants not only as culinary spaces but as semiotic, affective and ideological sites where diasporic identities are materialised and contested. Chineseness is understood as a historically produced, heterogeneous and continually negotiated diasporic imaginary.
The study combines ethnographic fieldwork, photographic documentation, Google Maps–based spatial mapping and interviews with restaurant owners, residents, and council officials, providing a multimodal account of how Chinese restaurants operate as cultural and ideological artefacts in a superdiverse suburb.
Using the Ethnographic Linguistic Landscape Analysis (ELLA) framework, the analysis is organised around three temporal dimensions. The present, through spatial–semiotic and onomastic analysis of shopfronts, shows how materiality and linguistic hierarchies shape diasporic visibility. The past, through diachronic comparisons including Google Street View archives and a mini ELLA case study, reveals how continuity and change are sedimented in the landscape. The future, approached as the proleptic orientation of semiotic practices and explored through affective regimes of nostalgia, insecurity and curiosity, shows how Chineseness is projected to imagined audiences and continually remade.
Theoretically informed by geosemiotics, materialist semiotics and affect in linguistic landscapes, the thesis conceptualises Chineseness as a chronotopic assemblage where historicity, everyday negotiation and futurity converge. It further reframes Sinoburbia not as static enclaves but as emergent ethnic spaces where Chinese restaurants operate as key semiotic and affective sites that produce evolving forms of identity, belonging and multiculturalism
Advances in Imperfect Supervision: From Multiple Unlabeled Sets to Weakly-Annotated Graphs
No full textSupervised machine learning has been a major driver of progress in artificial intelligence, powering applications across domains such as healthcare, robotics, and related fields. The methods from supervised learning typically rely on large datasets with accurate labels. However, in real-world settings, such perfectly labeled data is often unrealistic due to imperfections in data collection, including limited availability, missing values, and annotation errors. These challenges have led to the development of reliable and robust approaches that can effectively handle imperfect supervision, which commonly arises in three forms: inexact, incomplete, and inaccurate supervision.
This thesis investigates advanced topics spanning these three core forms of imperfect supervision. For inexact supervision, we introduce a novel problem setting for binary classification using multiple unlabeled datasets, which relies on minimal and easily obtainable supervision signals. In the context of incomplete supervision, we focus on graph-based positive-unlabeled (PU) learning and reveal how the structural characteristics of graphs can violate key assumptions of conventional PU approaches. Under inaccurate supervision, we tackle the problem of label noise in graph data by proposing a topological sample selection approach that leverages graph structure to identify clean and informative nodes more effectively.
Together, this thesis advances the understanding and capability of machine learning under imperfect supervision, particularly in structurally complex environments such as graphs. By systematically addressing challenges across inexact, incomplete, and inaccurate supervision, the proposed methodologies bridge theoretical principles with practical implementation and pave the way for more robust, adaptable, and trustworthy learning systems even when training data is coarse, partial, or noisy
Acute Effects of Novel Rest Interval Strategies in Resistance Training on Exercise -Related Outcomes Across Young, Older, and Clinical Cohorts
No full textResistance training (RT) is essential for improving neuromuscular function and functional capacity, yet the optimal manipulation of rest intervals (RI) remains poorly understood across diverse populations. Novel strategies, such as blood flow restriction during rest intervals (BFR-RI) and cluster sets (CS), have recently emerged in the literature. This thesis investigated the acute effects of BFR-RI and CS on various exercise-related outcomes in healthy young adults, older adults, and individuals with Charcot-Marie-Tooth disease (CMT), a population susceptible to fatigue and functional impairment.
Chapter 3 examined BFR-RI during high-load squats in trained young adults. BFR-RI produced similar mechanical, perceptual, and physiological responses to conventional RT, with a small but significant reduction in total repetitions, indicating slightly accelerated fatigue. Chapters 4–6 compared CS versus traditional sets (TRAD) during chest press (CP) and leg press (LP). Across populations, CS generally preserved mean concentric velocity (MCV), attenuated MCV loss - particularly in CP - and increased estimated repetitions to failure. Reductions in rating of perceived exertion were only observed in young adults, and lower-functioning CMT participants were more likely to terminate CP sets early. The study in young adults elucidates minimal sex and strength impact on MCV attenuation.
Overall, CS improved acute performance and mitigated neuromuscular fatigue, particularly during CP, whereas BFR-RI offered minimal advantage over conventional RT. These findings highlight RI as a key training variable and support CS as a practical strategy to maintain performance and potentially reduce neuromuscular fatigue in healthy and clinical populations. Future research should explore individualized CS protocols, alternative set structures, and long-term adaptations to optimize outcomes, safety, and adherence, especially in populations vulnerable to fatigue or functional limitations
Standard-dose PET(SPET) Images Synthesizing
No full textPositron emission tomography (PET) is a functional imaging modality that uses a radioactive tracer to visualize and quantify metabolic processes in the human body. It has demonstrated considerable clinical value in oncology neuropsychiatry and cardiology showing substantial promise for advancing cancer diagnosis and management cardiac care and surgery as well as neurological and psychiatric applications. However repeated imaging poses health risks due to cumulative radiation exposure. To address this critical issue there have been attempts to lower the injected activity to get low-dose PET (LPET) images. Because LPET involves less radiotracer accumulation than standard-dose PET (SPET) the reconstructions are noisier and can compromise diagnostic utility. To restore image quality recent work employs deep-learning methods to denoise LPET and synthesize images that closely approximate SPET quality. This thesis presents new deep learning methods for PET synthesis. First to counter the scarcity of paired LPET–SPET data we propose a data-augmentation pipeline that synthesizes LPET from existing SPET scans. Second we introduce a time-controlled model that encodes noise level as a timestep variable. Trained only on the lowest-dose data the model generalizes to unseen dose conditions (e.g. 50 10 1 ). Both methods were trained on a public dataset called ultra-low-to-high PET dataset. A subset containing PET images scanned by a Siemens scanner was selected with each subject having seven different dose levels. The augmentation strategy boosts performance over training with smaller datasets and works with several SOTA synthesis networks. The time-controlled approach outperforms strong baselines and remains robust when tested on dose levels withheld during training highlighting flexibility and wide generalization
Counselling in a cross-cultural context: The case of refugees from the Great lakes Region of Africa Who now live in Australia
No full textThis thesis investigates the psychosocial support needs of African Australians from the Great Lakes Region War Survivors (AAGLRWS), who resettled in Australia after enduring war, persecution, and displacement. It argues that culturally sensitive support must address trauma experienced across pre-migration, migration, and post-settlement phases. The study identifies structural and operational gaps in counselling and psychosocial services, underscoring the need for trauma-informed, culturally responsive care.
Using qualitative inquiry, the research explores the psychological and social suffering of 55 individuals from the Democratic Republic of Congo, Burundi, and Rwanda—countries shaped by colonialism, civil conflict, and human rights violations. Semi-structured interviews across four Australian states were analysed using grounded theory and narrative inquiry, informed by phenomenology.
Findings show that resettlement, influenced by push, pull, and residual factors, negatively impacted participants’ mental health, contributing to marginalisation, racism, and institutional neglect. Many expressed dissatisfaction with Western counselling models, which they viewed as culturally misaligned and ineffective in fostering recovery, agency, and integration. Nonetheless, some participants demonstrated resilience through entrepreneurship and the Ubuntu collective spirit.
The study highlights cultural mismatches in cross-cultural psychosocial support, particularly between individualistic and collectivist frameworks. It proposes a holistic, culturally integrated model incorporating Ubuntu and hybrid approaches tailored to AAGLRWS contexts. These findings inform the redesign of psychosocial services for AAGLRWS and offer broader relevance for supporting culturally and linguistically diverse (CALD) communities in Australia
The Accompanist as Dramaturg: An Exploration of Dramatic Function in Secco Recitative Accompaniment
No full textSecco recitative accompaniment, though ubiquitous in eighteenth-century opera repertory and central to modern operatic practice, has attracted surprisingly little systematic scholarship. This dissertation reconceptualises secco recitative accompaniment as live dramaturgical action, contending that the keyboard accompanist is not a neutral technician but a co-author of the performance text. Drawing on eighteenth-century theory (Sulzer, Scheibe, Quantz), theatre semiotics (Elam, Fischer-Lichte, De Marinis), Beckerman’s model of dramatic action, and Robert Hatten’s theory of virtual agency, it advances a structural–ontological framework that situates musical gesture, harmonic force, and rhetorical text-setting within a single field of performative praxis. As part of this framework, the study articulates the structural and expressive relationship between the embodied vocal line and its instrumental accompaniment, defining the interdependence through which recitative’s dramaturgical force emerges.
Methodologically the thesis proceeds in two parts. Chapters 2–5 synthesise historical treatises with contemporary dramaturgical theory, uncovering recurring principles of unity, variety, and gestural markedness that underpin effective recitative. Chapter 6 and Appendix 1 demonstrate these principles in practice through a beat-by-beat analysis of selected scenes from the Mozart–Da Ponte operas, applying a three-stage workflow that integrates dramatic segmentation, poetic–rhetorical scansion, and harmonic mapping, with analysis of applied gestural structure. The resulting framework furnishes a structured and adaptable tool for today’s continuo performers, enabling them to translate theoretical insights into coherent preparatory rehearsal procedures. It also establishes a conceptual foundation for understanding recitative accompaniment as a theatrical and performative entity, defining structural and expressive parameters that support future analytical and practice-led research. By bridging the long-standing gap between historical commentary and contemporary production, the dissertation empowers accompanists, singers, directors, and conductors to heighten narrative clarity and expressive depth in performance. Beyond its immediate practical value, the study opens new avenues for research into the dramaturgical agency of accompaniment in opera and related genres
Thermophoretic fabrication of gradient hydrogels for mechanobiology applications
No full textCell behavior during development, disease, and homeostasis are regulated by spatially dynamic cues found within their microenvironment, often in the form of stiffness gradients. In the context of mechanobiology, microengineered stiffness gradient hydrogels offer a powerful tool to probe how cells sense and respond to biophysical cues, particularly towards the identification of new mechanistic understanding and therapeutic strategies. However, the need to spatially manipulate the properties of soft hydrogels at the micron scale remains a major fabrication challenge. Therefore, this thesis aims to enhance the accessibility and adoption of gradient hydrogel technology in mechanobiology research, particularly by leveraging temperature-driven thermophoresis phenomenon and microfluidics technology to create a robust fabrication platform. Through extensive optimization and characterization studies, this work demonstrates four key features of the engineered platform that overcome several limitations of existing methods, including 1) highly precise and reproducible patterning, 2) compatibility with a wide range of hydrogel chemistries, 3) broad biologically relevant stiffness gradient regimes, and 4) the ability to create complex gradient patterns. Alongside this is the development of a new class of fluorescently labelled gradient hydrogels which display a stiffness-dependent fluorescence readout. This strategy enables quantitative assessment of the gradient formation process and contactless stiffness mapping via standard microscopy imaging, offering a simpler alternative to the gold standard atomic force microscopy for material characterization. Overall, the technical developments and findings gained from the series of cell-material interaction studies in this thesis have implications for advancing biomaterials technologies in mechanobiology, with potential impacts on cell signaling and how to modulate cell behavior by programming the microenvironment
Submission to the 2025 Review of the Australian Code of Practice on Disinformation and Misinformation
No full textThis submission proposes a reconceptualisation of the Australian Code of Practice on Disinformation and Misinformation. We argue that the current approach conflates two distinct types of harm requiring different policy responses: individual harm from exposure to dangerous content, and collective harm from systemic degradation of information quality across the digital ecosystem. We propose that platforms should be held accountable for preventing individual harm through content moderation, while contributing to ecosystem-wide monitoring of information disorder through a novel persona-based measurement system that protects epistemic rights
Analysis of Risk levels and use of AI in Automated Decision-Making systems in NSW
No full textDataset of human evaluations of risk level and use of AI. Using data from on https://cmsassets.ombo.nsw.gov.au/assets/Reports/Compendium-of-ADM-Systems.pdf. Methodological explanations in article ARE WE REGULATING THE RIGHT DIGITAL SYSTEMS? TESTING EMERGING ARTIFICIAL INTELLIGENCE FRAMEWORKS AGAINST REAL-WORLD PUBLIC SECTOR SYSTEMS - JOSE-MIGUEL BELLO Y VILLARINO,* KIMBERLEE WEATHERALL,** TERRY CARNEY,*** ALEXANDRA SINCLAIR, **** AND SCARLET WILCOCK in UNSW Law Journal Vol 48, issue 4.As describe
Quantifying dynamical properties of brain activity using complex systems analysis
No full textThe human brain exhibits a complex and multiscale dynamical structure, from microscopic cellular
crosstalk to macroscopic networks.
While complex systems science offers a plethora of analytical tools, brain dynamics are typically
explored with only a few hand-picked statistics.
This thesis quantifies the brain's intricate dynamical structure by integrating methods from complex
systems analysis into a systematic and interpretable framework for comparing across measures
sensitive to different aspects of neural activity.
Chapter 2 introduces time-series feature analysis through the lens of information theory, bridging
distinct measures with common notation and terminology.
Chapter 3 introduces and applies the Python library, pyspi, to time-series classification problems
(including neuroimaging datasets) to uncover the most salient features for a given task.
Building on this foundation, Chapters 4--7 apply these highly comparative methods to compelling
questions in modern neuroscience.
Chapter 4 comprehensively evaluates measures of inter-areal coupling to characterize its potential
role in conscious visual perception.
Chapter 5 expands the scope to integrate local activity and pairwise interactions in studying
neuropsychiatric disorders, supporting the continued use of linear measures while underscoring the
importance of multiscale approaches.
Chapter 6 probes functional, structural, and molecular correlates of homotopic connectivity, a robust
property of inter-hemispheric network architecture.
Finally, Chapter 7 extends this framework to systematically compare algorithms that capture
overlapping communities in the structural connectome to test new biological hypotheses.
Collectively, this thesis presents a highly comparative framework for capturing the complex and
multiscale dynamical structure of the human brain.
This work points to exciting future directions in fields from lifespan development to personalized
medicine in an era of expanding openly available data