21684 research outputs found
Sort by
Place attachment and urban change: Developing a phenomenological methodology for studying the lived experience of affective bonds to place
No full textAbstract not currently available
The effectiveness of simulation-based learning in nursing education
No full textAbstract not currently available
Development and testing of a MEMS based gravimeter for field survey use and airborne deployments
No full textAbstract not currently available
Nanopatterning titanium and PEEK for orthopaedic implants
Get PDFThis project was inspired by research by Dalby and Gadegaard that demonstrated nanopatterning of poly-methyl-methacrylate (PMMA) surfaces can stimulate mesenchymal stromal cells (MSCs) to differentiate into osteoblasts and produce bone mineral in vitro.[1] The motivation for this thesis was to adapt and upscale the technology for clinical application, with the aim of fabricating osteogenic imlants for orthopaedic surgery, such as intervertebral fusion cages.[2] This translation would initially involve injection mould nanopatterning poly-ether-ether-ketone (PEEK) surfaces. A further objective was to discover methods for fabricating non-planar moulds that could be used in the injection mould nanopatterning process.
Nanoimprint lithography of a novel titanium dioxide precursor sol-gel was performed using flexible polydimethylsiloxane (PDMS) stamps that could conform to non-planar contours of injection mould inlays as a demonstration of the technology. Subsequent injection moulding showed initial success, but the titanium dioxide nanopillars lacked the durability required for repeated moulding cycles.
Nanopatterned PEEK surfaces produced by injection moulding (using electroplated nickel inlays) were assessed to determine whether the nanopatterns exhibited any biological effect upon human bone marrow cells. Initial in vitro experiments by Dr Daniel Morrison and a collaborative group in Davos raised concerns regarding cell adhesion on nanopatterned PEEK surfaces and additional work was undertaken to modify PEEK using oxygen plasma treatment.[3] The use of a cell seeding device designed by Dr Paul Reynolds, led to more reliable in vitro results as it provided a more favourable environment for cell adhesion.
Due to the opacity and autofluorescence of PEEK, in vitro analysis used histological staining with reflected light microscopy and quantitative reverse transcriptase PCR. In vitro experimentation revealed that oxygen plasma treatment increased cell adhesion but reduced the bioactive effect of nanopatterning. Although bone marrow cells adhered to the PEEK nanopatterns in small numbers, the cells exhibited a more osteogenic phenotype, demonstrated by relative increased in calcium and phosphate expression.
Nanopatterned PEEK did not achieve the results required for progression to an in vivo study. Therefore, surface coating nanopatterned PEEK was considered as an alternative method to satisfy the objectives of the project. An in vivo study was undertaken in collaboration with Nijmegen to study osseointegration of titanium coated injection mould nanopatterned surfaces. Due to intellectual property negotiations, polycarbonate was used rather than PEEK and the NSQ and HEX nanopatterns were not included. The titanium coated nanopatterned implants demonstrated significantly increased bone to implant contact compared to commercially developed grit-blasted acid-etched titanium implants.
With a view to further pre-clinical studies of nanopatterned implants, improved in vivo models of osseointegration and osteogenesis in rabbits were developed. These will enable the assessment of novel implants and satisfied the UK Home Office requirements for reduction, refinement and replacement of animal models.
Although not suitable for use in high performance injection mould inlays, the titanium dioxide precursor sol-gel developed for this thesis could be used to directly nanopattern orthopaedic implant surfaces, thus promoting osteogenesis. Furthermore, as demonstrated by the in vivo study presented in this thesis, injection mould nanopatterned polymeric implants (such as PEEK) can be modified with an ultra-thin layer of titanium to improve osseointegration.
The work described herein has highlighted that nanopatterning will not necessarily provide the same results in different materials. It does, however, provide further evidence to support the hypothesis that nanopatterning directs cell behaviour by nanotopographical changes in surface chemistry and surface energy which affect cell adhesion
Censorship and female identity in contemporary China
Get PDFThis dissertation examines how theatre censorship operates in Xi Jinping's China, particularly in feminist theatre. By focusing on interviews with Zhiheshe members, a student organization that performed The Vagina Monologues for over a decade, this research highlights the strategies feminist practitioners adopt to navigate institutional, public, and self-imposed censorship. Drawing on theories of power by Foucault, Butler, and Bourdieu, it explores how censorship not only restricts feminist voices but shapes public discourse on gender and identity. The study fills a gap in literature on contemporary feminist theatre in China, providing insight into the complex dynamics between censorship, gender, and cultural expression in an authoritarian context
Architectures and optimisations for FPGA-based simulation of quantum circuits
Get PDFThe increasing complexity and scale of quantum algorithms, coupled with the current limitations of physical quantum hardware, have led to a growing need for efficient quantum circuit simulation techniques. While CPUs and GPUs have traditionally been used for simulating quantum circuits, their energy consumption and scalability issues have prompted exploration into alternative platforms. Field-Programmable Gate Arrays present a promising alternative, offering the potential for customisable parallelism, energy efficiency, and flexible hardware configurations. This thesis investigates the use of FPGA architectures for Full State Vector Quantum Circuit Simulation, evaluating their performance, scalability, and energy efficiency relative to traditional CPU and GPU platforms.
The core aim of this work is to explore whether scalable FPGA architectures can be designed for quantum circuit simulation, and to assess their comparative performance and energy efficiency against established CPU and GPU solutions. The work was guided by several research questions: Can FPGA architectures be optimised for quantum circuit simulation? How does the performance of FPGA architectures scale with hardware utilisation? What types of circuits benefit most from FPGA-based simulation? Is there a performance-per-Watt advantage to using FPGAs over GPUs and CPUs?
To answer these questions, a variety of FPGA-based architectures were designed and evaluated. The architectural approaches investigated include Direct Iteration Processing, Buffered Architectures, and Gate Fusion Architectures. Each of these architectures was tested on benchmark quantum circuits, including Quantum Fourier Transform, and Grover’s search algorithm, representing a range of qubit counts and gate complexities. These architectures were compared in terms of scalability, execution time, and energy consumption, with their performance assessed against CPU and GPU implementations. One of the key contributions of this thesis is a controlled gate scheduling optimisation designed to improve performance for control-heavy circuits (i.e. circuits with a high number of controlled and multi-controlled gates). This architecture demonstrated substantial performance improvements for some circuits, where it was up to 5× faster than the baseline architecture. While the GPU still outperformed the FPGA in raw speed, the optimised architecture showed a significant energy advantage, consuming 2.6× less energy than the GPU for circuits with a high density of controlled gates. This highlights the potential of FPGA architectures to outperform traditional platforms in energy-constrained environments.
This work also presents a set of circuit width reduction techniques aimed at improving the scalability and efficiency of quantum circuit simulations on FPGA hardware. These techniques reduce the number of qubits required by identifying and transforming portions of the circuit that can be simplified without affecting the overall computation. Initially developed for circuits defining algorithms employing computational basis data encoding, the techniques were extended to handle circuits implementing algorithms employing the more widely-used amplitude-based data encoding approach, demonstrating their versatility. These optimisations were applied to circuits for computational fluid dynamics and quantum arithmetic, leading to more efficient use of FPGA memory and computational resources.
The introduced FPGA-based quantum circuit simulation platform is, to our knowledge, the first of its kind capable of simulating general-purpose quantum circuits, rather than being limited to specific algorithms or gate sets. Unlike many existing FPGA simulators that are specialised for particular quantum algorithms, such as Grover’s search or the Quantum Fourier Transform, this platform is designed to simulate any quantum circuit regardless of its structure or gate complexity, at high numbers of qubits (> 25). We simulate general-purpose quantum circuits of up to 29 qubits in this work, but in theory, the platform can scale up to any number of qubits, given sufficient memory resources. This level of flexibility, combined with the ability to handle larger quantum systems, positions this platform as a significant step forward in FPGA-based quantum circuit simulation, making it a versatile and scalable tool for both research and practical quantum computing applications.
Overall, this thesis demonstrates that while FPGAs may not match the raw execution speed of GPUs, they offer significant advantages in terms of energy efficiency for quantum circuit simulation, particularly for control-heavy circuits. The control scheduling optimisation and buffering strategies were found to significantly improve performance, especially for circuits with high controlled-gate density. However, challenges remain in terms of scalability, with High-Level Synthesis limitations posing barriers to further performance gains. The use of multi-FPGA clusters and further advancements in High-Level Synthesis tools could address these limitations and enable FPGAs to handle larger quantum circuits more efficiently
Enforcement, extension, and entanglement: Punishing non-payment of financial penalties in Scotland
Get PDFTaking the neglect of fines and financial penalties within contemporary criminological research as its starting point, this thesis examines fines enforcement as a distinct part of financial punishment processes in contemporary Scottish criminal legal practice. Using empirical evidence from 9 qualitative interviews with practitioners working in community justice in Scotland, this study offers evidence of the processes, practices, and outcomes of fines enforcement action.
The findings indicate that, for those who cannot pay, fines enforcement can involve an extended entanglement with the criminal legal system that can produce significant barriers to progress for those subject to fines enforcement action and/or living with unpaid financial penalties. Contextualising these findings within 40 years of policy and practice developments in Scotland, this thesis demonstrates how different and conflicting principles concerning the nature and purpose of non-custodial alternatives in Scotland have produced a fractured subfield where fines enforcement appears to undermine the goals of community justice.
This project contributes to a wider international resurgence in critical criminological research concerning of fines, financial penalties, and financial punishment, providing an upto-date account of practice in Scotland and demonstrating how findings can be used to inform theorisation about the role of fines enforcement in contemporary legal systems – especially in systems where decarceration and progressive alternatives are promoted as central and formative principles of criminal legal practice
Effective and efficient transformer models for sequential recommendation
Get PDFIn the last decade, advances in natural language processing have driven significant interest in Deep Learning-based Sequential Recommendation Systems, as user-item interaction sequences resemble word sequences in language models. In particular, the arrival of the Transformer architecture transformed the field of sequential recommendation. It allowed Transformer-based models, such as BERT4Rec and SASRec, to achieve state-of-the-art results on many sequential recommendation problems. However, while these Transformer-based models perform well on small-scale academic datasets, they face challenges in real-life applications due to scalability problems and the complexity of modern recommendation goals, which include beyond-accuracy goals such as recommendation diversity. In this thesis, we closely examine the sources of these le solutions to enable Transformer-based models for large-scale, real-world deployments.
In particular, training sequential recommenders is problematic. Indeed, most recommendation datasets contain different sets of items, making the pre-training foundation models impossible and requiring training recommendation models from scratch for every new recommendation dataset. Long training is problematic because it increases running costs and causes delays in fresh data processing. In our reproducibility study, we find that practitioners often end up with underfit models due to the long training requirement. To tackle the long training problem, we propose Recency Sampling of Sequences (RSS), a novel training objective for sequential recommender systems that allows the achievement of strong results even when training time is limited. For example, on the MovieLens-20M dataset, RSS applied to the SASRec model can result in a 60% improvement in NDCG over a vanilla SASRec and a 16% improvement over a fully trained BERT4Rec model despite taking 93% less training time than BERT4Rec.
Another big challenge for Transformer-based Sequential Recommender Systems is a large catalogue of items that may be several orders of magnitudes larger when compared to the vocabularies of items. Large catalogues create the need for negative sampling during training, but in this thesis, we show that negative sampling causes effectiveness degradation. To mitigate this problem, we design a new gBCE loss, which counters the effects of negative sampling by down-weighting the contribution of the positive sample in the overall cost. We show that gBCE allows for state-of-the-art effectiveness with large catalogues, even with retaining negative sampling.
A large catalogue also makes the item embedding tensor large and model inference slow, as sequence embedding is multiplied by this large embedding tensor. On the large-scale Gowalla dataset, where training non-sampled models is infeasible due to large catalogue size, we obtain substantial improvements by enhancing SASRec with gBCE loss (+47%). We also reduce the memory footprint and speed up model inference using our proposed RecJPQ technique that atomic item IDs into compact compositional sub-item ID representation.
Building upon RecJPQ’s sub-item representations, we also address the problem of slow model inference with large catalogues. In particular, we propose two algorithms for fast item scoring. First, we propose the PQTopK algorithm, which computes item scores as the sum of the sub-item scores. Sub-item scores can be pre-computed and re-used between items, which results in up to 4.5× faster item scoring when compared to regular Transformer’s scoring. We further observe similarities between RecJPQ sub-item representation and bag-of words representations in Information Retrieval (IR). In IR, the problem of fast-scoring large collections of documents has been addressed using Dynamic Pruning approaches that allow finding Top-K items without scoring the whole catalogue exhaustively. Building upon the similarities between item representations in RecJPQ and document representations in IR, we propose the RecJPQPrune dynamic pruning algorithm for the RecJPQ-based recommenders. RecJPQPrune further improves scoring up time to 5.3× compared to PQTopK and up to 64× compared to regular Transformer’s scoring.
Finally, while existing Transformer-based models perform well when measured using accuracy-based ranking metrics (e.g. NDCG), they usually struggle to optimise more complex goals, such as increasing diversity or promoting popularity bias. To improve model effectiveness on these complex beyond-accuracy goals, we propose an autoregressive Next-K recommendation strategy as an alternative to the traditional ”score-and-rank approach”. We also propose a universal reinforcement learning-based alignment scheme for the Next-K strategy and show that it is possible to align a generative recommendation model with beyond-accuracy goals, such as diversity promotion. Our experiments on two datasets show that in 3 out of 4 cases, GPTRec’s Next-K generation approach offers a better tradeoff between accuracy and secondary metrics than classic greedy re-ranking techniques for diversity optimisation and decreasing popularity bias
A critical analysis of decolonisation of postsecondary education in Canada
Get PDFIn 2015, Canada’s Truth and Reconciliation Commission of Canada (2015b) released a report entitled Calls to Action. This report provided clear steps for Canada to follow to atone for the cultural genocide committed against Indigenous Peoples through the Indian Act (1876) and in particular the Residential School system. Several of the calls to action focus on the education system. Since the release of this report, indigenisation has become a focus of Canadian postsecondary education strategic plans. My dissertation sought to better understand what is meant by the term indigenisation, which political, economic, and structural forces are impacting indigenisation, and how indigenisation efforts are or are not meaningfully moving decolonisation forward in Canada. As a settler in Canada who spent my formative years living and going to school alongside Indigenous youth and then lived and raised my children in a remote Inuit community, my goal was to be able to illustrate what the ideal future could look like where indigenisation policies were effectively contributing to decolonisation.
Deconstructing the current state required an examination of past policy and discourse to understand the systemic factors competing with indigenisation. I first explored key theoretical constructs: poststructuralism, postmodernism, critical race theory, colonialism, postcolonialism, decolonisation, whiteness, neoliberalism, and social justice. After discussing my methodology and the postsecondary education system in Canada, I examined the Indian Act (1876), the results of Aboriginal Commissions in the 1990s, the Truth and Reconciliation Commission of Canada’s (2015a, 2015b) work, and the impetus to embed indigenisation within public and postsecondary education. After describing the current system of postsecondary education in Canada and the political, economic, and structural drivers for indigenisation, I shared the results of my examination of college websites, as I sought to understand how ingrained the indigenisation efforts were.
The latter part of the dissertation focused on the meaning of indigenisation, what it looks like in its current state, and how it could look in an ideal future. Recognising the clash between neoliberalism and social justice movements, and the human need to categorise leading to “otherness,” I suggested that decolonising requires a disruption and dismantling of the current postsecondary education system.
This dissertation contributes to the discussion on how to dismantle colonial institutions such as postsecondary education. It explains the need to move beyond token actions, beyond inclusion practices, beyond hiring some Indigenous staff and professors to carry the workload of indigenising the academy. I argue that creating the third space that Bhabha (1994) referred to requires ceding power and rebuilding a new framework of governance. This requires reimagining and reconstructing the future of education for Canada in whichbinaries no longer exist and change happens at all levels. Then, and only then, can Canadians say that reconciliation has taken place and decolonisation is underway
Data sources:
Access Repository Dashboard
Do you manage Open Research Online? Become a CORE Member to access insider analytics, issue reports and manage access to outputs from your repository in the CORE Repository Dashboard! 👇