16168 research outputs found
Sort by
Bicycles of perception: Motivations and meanings that promote long-term engagement with mountain biking in Queenstown, Aotearoa New Zealand
No full textThis research explores the motivations and meanings of mountain bikers living in the Queenstown Lake District, Aotearoa New Zealand. Through a phenomenological paradigm, this research follows a qualitative methodology. Data for this research was gained through an ethnographic process, involving interviews, observations, and a self-practice of mountain biking. Semi-structured interviews were used to gain insight into mountain bikers' personal experiences. Observations were conducted to examine numerous aspects of mountain bikers' lifestyles. Observations were conducted at mountain bike tracks, mountain biking events, and social events; e.g. barbecues and bars. Self-practice by myself, the researcher, was used as an attempt to gain further insight into the subjective experiences of embodied action. The data was analysed via grounded theory to allow themes to emerge. Numerous themes came to light during data analysis. They reveal some of the ways mountain biking affects participants' lives. Some of the prominent themes to emerge were physical and mental health, connection to nature, stress relief, social connection and identity. The state of flow and deep embodied action mountain bikers experience, appears to promote long-term engagement. The above themes are strongly associated with overall wellbeing. They indicate that mountain biking offers a holistic approach to maintaining a well-balanced life. This research will give some insight into why mountain bikers continue to participate in a risky and physically demanding sport over long periods of time
Building trustworthy smart contracts using interactive theorem proving
No full textThere are varying approaches to the verification of smart contracts using formal methods. This thesis advocates for the use of high-level specifications coupled with a verified compiler to low-level bytecode, such as for the Ethereum Virtual Machine. Taking this approach allows for specifications to more closely match natural language, while ensuring that the specifications apply to the real bytecode executed on-chain.
Interactive theorem proving can provide the foundation for developing provably correct smart contracts. Due to the immutable nature of smart contracts and their potential to manage highly valuable assets and tokens representing power, techniques to ensure their correctness are of paramount importance.
This thesis extends the DeepSEA (Deep Simulation of Executable Abstractions) smart contract language targeting the Ethereum Virtual Machine by mitigating the issues associated with reentrancy and introducing a model of relevant aspects of a blockchain. This enables the specification and verification of two case studies which exemplify the approach of developing provably correct smart contracts.
The specifications for the case studies are written in the language of the Coq Proof Assistant, making arbitrary mathematical statements expressible. The blockchain model enables stating and proving temporal properties relating to the execution of smart contract over time.
While smart contracts are an ideal application area for formal methods in general and interactive theorem proving in particular, the techniques exemplified in this thesis could be applied throughout software engineering. The relatively young age of smart contract languages and typically small size of smart contract programs makes the application of interactive theorem proving more tractable. Future work could involve demonstrating the applicability to many interrelated smart contracts and to larger software projects in different domains.
The first three chapters of this thesis cover introductory and background material. This is followed by the contributions to the DeepSEA system. The two case studies and the proof themes arising from them are then presented. Finally, concluding remarks and future work are discussed
Eliciting Expert Uncertainty from Decision Making
Get PDFEliciting expert uncertainty is a complex task with many caveats. Its usefulness in high-level decision-making and Bayesian inference makes it a necessary task that must be completed accurately. This thesis explores uncertainty elicitation in terms of prior elicitation for Bayesian inference and the wider knowledge elicitation field. It describes methods currently available for prior elicitation and proposes a new typology, highlighting lesser-known methods. Based on the limitations of current methods, a new method for prior elicitation is introduced. This new method allows an analyst to model expert decision-making data to elicit a probability distribution that reflects expert uncertainty. An example of parole board decision-making is used to elicit a prior distribution of a prisoner re-offending upon release from prison. This example shows analysts how to elicit distributions for tabular data; however, more complex data types, such as images and reports, are often used for decision-making. To elicit distributions capturing expert uncertainty from more complex data, this thesis introduces a deep learning approach and uses an example of eliciting cancer risk, in histopathology, to illustrate this approach for the wider knowledge elicitation field
Effects of raised safety platforms (RSPs) on travel time and traffic delay at roundabouts
No full textRaised safety platforms (RSPs) are vertical deflection devices that utilise vertical acceleration to slow vehicle operating speeds down to safe systems-compliant speeds. These devices are not as jarring for drivers as a traditional judder bar and are specifically designed for motorist comfort when traveling at a safe and appropriate speed. RSPs have been specifically identified through the Austroads Guide to Traffic Management and New Zealand Transport Agency (NZTA) Standard Safety Intervention Toolkit as a key safe systems safety intervention.
RSPs are also part of best practice guidance and are designed to slow vehicle speeds enough so that when people make mistakes, they have time to react and avoid a crash. If a crash were still to happen it is kinetic impact energy which determines the severity outcome of a crash is lower. At lower speeds the likelihood of a death or serious injury occurring as a result is drastically reduced.
Safe Systems is a road safety philosophy that requires industry practitioners to provide a more forgiving road system reducing the harm to the human body in the event of a crash. Safe systems have four key pillars safe roads and roadside, safe speeds, safe vehicles, and safe road users. Safe speeds are the areas within which RSPs sit.
Although there are proven safety benefits of RSPs through various literature sources there is still a common user perception that these devices cause a decrease in operational efficiency. Literature is also light on this topic and in cases where past literature is available, they seem to contradict with each other. Some user arguments are that RSPs create congestion at intersections resulting in an increased travel time.
The aim of this thesis report was to provide data-based clarity on the effects of the RSP on travel time and traffic delay at roundabouts to help clarify this area of knowledge gap. This report also seeks to understand effects on traffic speed, crashes, queue lengths, and LOS at roundabouts as a result of RSPs.
This report specifically focuses on roundabouts due to a lack of roundabout-related operational literature as the majority of RSP applications have been at signalised intersections. It should be noted that RSPs are still a new and emerging idea for Australia and New Zealand (NZ). RSPs are a widely used safety intervention in European countries such as the Netherlands.
Investigation in this paper has been done through a review of the literature and quantitative assessment of a sample of sites through microsimulation traffic modeling using Multimodal Traffic Simulation Software. Based on the findings of this paper it is identified that there is a negligible change in traffic delay and a decrease in travel time as a result of RSPs at roundabouts.
The findings of this research shows that RSPs create smoother traffic flow, and safer and better gap selection for vehicles entering a roundabout resulting in improved or the same efficiency for the intersection when compared to pre RSP operation. Some increases in queue lengths was evident on the approach to RSPs and this could relate to driver perception of an increase in congestion. There is no evidence from this paper that supports this as the overall LOS remains unchanged or is improved.
This paper also found that speeds dropped notably on approach to RSPs which resulted in a change in driver psychology where drivers were yielding more at the slower speed. Crash data also showed a decreasing trend in the number of crashes and their severity following RSP installation across all the sites studied. These finding also support past literature on RSP speed and crash reductions
X-ray diffraction for phase identification in Ti-based alloys: Benefits and limitations
No full textX-ray diffraction (XRD) is routinely used to characterise Ti alloys, as it provides insight on structure-related aspects. However, there are no dedicated reports on its accuracy are available. To fill this gap, this work aims at examining the benefits and limitations of XRD analysis for phase identification in Ti-based alloys. It is worth mentioning that this study analyses both standard and experimental Ti alloys but the scope is primarily on alloys slow cooled from high temperature, thus characterised by equilibrium microstructures. To be comprehensive, this study considers the all spectrum of Ti alloys, ranging from alpha to beta Ti alloys. It is found that successful identification and quantification of the phases is achieved in the majority of the different type of Ti-based alloys. However, in some instances like for near-alpha alloys, the output of XRD analysis needs to be complemented with other characterisation techniques such as microscopy to be able to fully characterise the material. The correlation between the results of XRD analysis and the molybdenum equivalent parameter (MoE), which is widely used to design Ti alloys, was also investigated using structural-analytical models. The parallel model is found to be the best to estimate the amount of β-Ti phase as a function of the MoE parameter
Self-supervised Feature Extractor Training for Alzheimer’s Disease Classification
Get PDFDeep learning achieves encouraging performance in natural image classification. It has huge potential for detecting neural degenerative anomalies, but this is often limited by the availability of well-segmented neuroimages and computational
resources. One way to address this problem is to apply self-supervised learning methods that utilise unsegmented neuroimages and artificial labels. Another solution is developing surrogate tasks to learn the representations of neuroimages for classification. This thesis reports the investigations of different variants of self-supervised learning and pretext tasks to train feature extractors for downstream Alzheimer’s Disease classification.
Firstly, this thesis reviews the literature regarding Alzheimer’s Disease classification and possible data leakage issues. Then, a lightweight 3D CNN-based ensemble is trained to predict brain age using the 3D MRI data of cognitively normal subjects from the OASIS-3 dataset. The extracted features are evaluated in the binary classification of CN vs. AD patients from their brain MRI scans. This approach achieved competitive performance compared with state-of-the-art methods in the literature.
The next part of this thesis developed four different self-supervised learning pretext tasks for feature extractor training: brain age prediction, brain sMRI reconstruction, brain sMRI rotation classification, as well as a combination of all three approaches into one single multi-task predictor. To further explore the feasibility of employing synthetic neuroimaging data in the self-supervised learning setting, the proposed approaches are trained on the LDM100K dataset followed by evaluation using real-world OASIS and ADNI datasets. The real-world data training and testing leads to the best classification performance. The random cropping data augmentation technique can improve feature extractor training on 3D MRI data. Due to high computational expense and time limitations, the results of the training using synthetic data are not as satisfactory as those using real-world data. Future research is needed to develop more advanced feature extractor architectures and more complex pretext tasks that can learn more discriminative features. Another area of research to improve training efficiency would involve developing specialised software and hardware for processing 3D neuroimaging data
The integration of queer-related curriculum in psychology training in Aotearoa/New Zealand: A survey with programme directors
No full textAs a key profession within the mental health workforce, psychologists can offer affirmative services that lead to positive therapeutic outcomes for queer clients. This study examined how training programmes to become a registered psychologist in Aotearoa New Zealand currently provide content on working with queer clients. In 2022, all (N = 17) programme directors of psychology training were invited to participate in a survey that assessed the cultural responsiveness of their training programmes and 15 responded. One-fifth (19%) classified their programme as containing at least ‘a moderate amount’ of queer content. Four-fifths (79%) reported that knowledge of caring for queer clients would be ‘very’ or ‘extremely’ important in psychology practice and/or thought more time should be dedicated to such content. Over half (57%) adopted an ‘add on’ approach to deliver queer content (e.g. guest lectures). During a time when queer communities experience drastically high rates of mental health issues and high exposure to minority stressors, there is a critical gap in the meaningful threading of queer content across the psychology curriculum. We have outlined ten points for reflection on developing this aspect
Efficiency of cycled batteries analyzed through voltage-current phase differences
No full textAgeing of rechargeable batteries is routinely characterized in the frequency domain by electrochemical impedance spectroscopy, but the technique requires laboratory measurements to be made on a time scale of days. However, the normal cycling of a battery as it is used in situ provides equivalent information in the time domain, though extracting robust frequency information from a time series is challenging. In this work, we explore, in the time domain, the relationship between instantaneous voltage-current phase difference and cycle efficiency. Moreover, we demonstrate that phase measures can be used to identify battery ageing. We have cycled a 250 mA h Nickel-Cobalt cell several hundred times and used Hilbert Transforms to identify phase difference between voltage and current. This phase difference becomes closer to zero as the battery ages, commensurate with a drop in energy cycle efficiency. In another experiment, we applied a synthetic current profile mimicking behaviour of an electric car cell, to a 3.2 A h LiNiMnCoO2 cell, for 100 days. For this more complicated profile with a wide range of frequency content, we used wavelet analysis to identify changes in phase difference and impedance as the battery aged. For this cell, drop in cycle efficiency was associated with a rise in internal resistance. The results imply that time-series analysis of in situ measurements of voltage and current, when applied with equivalent circuit models and underlying theory, can identify markers of battery ageing
On D-inverse constellations: An alternative view of ordered groupoids
No full textA groupoid is a category in which every arrow has an inverse. The ESN (Ehresmann-Schein-Nambooripad) theorem states that the category of inverse semigroups is isomorphic to the category of inductive groupoids, which are groupoids with additional order-theoretic structure. Inductive groupoids are special types of ordered groupoids; the latter shares most of the properties of inductive groupoids but do not correspond to any type of semigroup. Despite this, many of the main facts about inverse semigroups carry over to ordered groupoids, and ordered groupoids have been shown to be an important tool in the study of inverse semigroups. Since then, it has been found that a particular type of partial algebra we call D-inverse constellations are equivalent to ordered groupoids, but have an arguably simpler, purely algebraic definition. This thesis will explore and illustrate the value of working with D-inverse constellations rather than ordered groupoids, presenting an alternative formulation and proof of the ESN Theorem using D-inverse constellations
The 'glow up' imperative: The fitness lifestyles of young women in Aotearoa New Zealand
No full textFor many young women, their relationships with fitness, health and the body are shaped by unrealistic beauty ideals. Body image scholars have long focused on the impacts of the media and more recently, social media, on the way young women experience body satisfaction. This thesis investigates the lived experiences of young women in Aotearoa New Zealand, in relation to body image and fitness culture. Underpinned by feminist post-structuralist theory, this research project aims to amplify the voices and experiences of young women (18-25 years) to gain nuanced insights into how they interpret, internalise and negotiate the complexities of the fitness lifestyles and everyday body image. This study draws upon an anonymous survey (203 participants), as well as three focus groups and three individual interviews with 15 young women, with all participants considering themselves to be actively pursuing fitness lifestyles.
Conducting a thematic analysis, results showed that many young women are actively pursuing a range of fitness practices for diverse reasons and motives. However, many report feeling the pressure not only to adhere to thin ideals, but also to pursue bodies that are strong and toned. For many of the young women, however, the fitness lifestyles were also connected to broader social trends of ‘fitspo’ (fitness inspiration) and the ‘glow up’ (body and lifestyle transformation). Engaging with feminist literature on the pressures on young women to aspire and achieve ‘successful’ femininities in the context of neoliberalism, this research develops the concept of the ‘glow up’ in relation to their highly consumptive fitness lifestyles. In so doing, this thesis provides rich, nuanced and original insights into the pressures on everyday young women to pursue bodily and lifestyle transformation, and the effects on those unable to maintain their commitment to the ‘glow up’ imperative. Importantly, not all young women passively accept such limiting versions of the ‘body project’, with some actively acknowledging the harm that can be caused by the ‘glow up’ imperative. The thesis concludes with some researcher reflections on the contributions of this study and future directions to help young women redefine what it means to participate in meaningful physical activity