Search

Statistical learning (SL), the ability to detect and extract regularities from inputs, has been considered as an early-maturing and domain-general mechanism that is critical for typical language development. However, recent evidence in neurotypical adults and children have found that individuals can vary in their SL abilities across linguistic and nonlinguistic domains as well as across auditory and visual modalities. Moreover, rather than being developmentally stable, SL was found to improve with age in the nonlinguistic domain, while mixed evidence has been found in the linguistic domain. A substantial gap still exists in our understanding of how SL in the linguistic domain changes across development and across individuals with different levels of language experience. This dissertation aimed to understand a). how SL develops across linguistic and nonlinguistic domains in neurotypical adults and children, b) whether SL can serve as a predictor of language development in children with language difficulties, c). how domain-general attention system is engaged during SL in typical development, and d). whether domain-general attention can facilitate SL in children with language difficulties. Chapter 2 to 3 revealed that neurotypical children demonstrated a strikingly faster learning rate and greater learning-induced neuroplasticity in the language network during linguistic SL when compared to neurotypical adults. Chapter 4 showed that autistic children who have heterogenous language skills exhibit specific impairment in linguistic SL, while their nonlinguistic SL was comparable to neurotypical children. A potential reciprocal relationship was found between linguistic SL and language development in autistic children. Chapter 2 - 4 highlighted that linguistic SL varies across development. The plasticity of the language network and the amount of accumulated language exposure of the learners could both determine the outcomes of linguistic SL. Given the unique role of linguistic SL in language development, it is important to understand factors that can facilitate linguistic SL across development. Attention is a late-maturing and domain-general cognitive function that supports learning across domains and development. However, clarifying the role of attention in SL remains an area of ongoing research. It remains unknown how attention is engaged during linguistic SL across development. To answer this question, Chapter 5 Study 1 showed that neurotypical children showed a significantly greater learning-induced change than adults in both the dorsal attention network and the language network during auditory linguistic SL. During the later stage of learning, neurotypical children also showed a significantly stronger change towards anticorrelation between the dorsal attention network and the language network compared to adults. These findings highlighted a changing relationship between the attention and language network during linguistic SL across development. The greater short-term plasticity observed in children compared to adults in both networks might confer an advantage for children in learning regularities from new linguistic inputs. Chapter 5 Study 2 further revealed that orienting children's attention to specific stimuli during the learning process facilitated linguistic SL across neurotypical children, autistic children, and non-autistic children with reading difficulties. The ability to learn statistical patterns from attended visual linguistic inputs predicted gains of emergent literacy skills in a reading intervention, and improved literacy skills, in turn, enhanced SL capacities in children with reading difficulties. These findings together suggested that through its interaction with both the language network and language experiences, attention plays a crucial role in modulating linguistic SL, which in turn, shapes language development across populations with varying neurodevelopmental trajectories. [The dissertation citations contained here are published with the permission of ProQuest LLC. Further reproduction is prohibited without permission. Copies of dissertations may be obtained by Telephone (800) 1-800-521-0600. Web page: http://www.proquest.com/en-US/products/dissertations/individuals.shtml.]

The capacity of educators to dynamically adjust their thought processes and behaviors in response to the diverse needs and characteristics of students in an unfolding teaching situation is widely acknowledged as a cornerstone of effective teaching. While teachers' instructional adaptations have been documented and studied from various angles, the field lacks a robust theoretical framework that precisely distinguishes yet cohesively integrates teachers' cognitive processes and capabilities of adaptive teaching. Addressing this gap, this dissertation presents the Teacher Adaptive-Cognition Theory (TACT), an innovative theoretical framework that aims at providing a comprehensive and much-needed perspective on the cognitive complexity inherent in adaptive teaching. Formulated through a critical and extensive review of adaptive cognition within and beyond the field of teaching, the TACT model distinguishes among behavioral performance of adaptation, cognitive performance of adaptation, and cognitive readiness for adaptation. It posits that analyses of the cognitive processes and capabilities of adaptive teaching should be systematically contextualized within each other. Furthermore, the dissertation delves into the implications of the TACT model for both research and practical application within and beyond the realm of teacher adaptability. By providing this detailed, integrative framework, this dissertation aims to enhance the conceptual clarity needed to guide research in the study of the crucial skill of teacher adaptability in particular and adaptive expertise in general, as well as to foster the development of such skill. [The dissertation citations contained here are published with the permission of ProQuest LLC. Further reproduction is prohibited without permission. Copies of dissertations may be obtained by Telephone (800) 1-800-521-0600. Web page: http://www.proquest.com/en-US/products/dissertations/individuals.shtml.]

Quantum mechanics is notoriously challenging, and research has found that students struggle with many common difficulties when learning it. It is also proving to be a critical piece of many exciting fields that are all but assured to see great development and expansion in the coming years; the Second Quantum Revolution is upon us. Quantum information science and engineering is a rapidly unfolding field, requiring talent from many disciplines, that aims to leverage the potential of quantum systems for many practical applications. To prepare students for the opportunities afforded by these advances, a strong foundation in quantum mechanics is essential. The work that I present in this dissertation is focused on helping students achieve this understanding. By investigating the common difficulties that students have in key concepts related to quantum mechanics and quantum computing, a guiding framework can be established and followed to develop research-validated, active-engagement instructional tools. These tools include Clicker Question Sequences (CQSs) on (1) the basics of two-state quantum systems, and changing basis in two-state systems; (2) time-development of two-state systems; (3) quantum measurement of two-state systems, and (4) measurement uncertainty in two-state systems. In addition to these, I have developed and validated Quantum Interactive Learning Tutorials (QuILTs) consisting of guided-inquiry teaching-learning sequences for (1) the Bloch sphere and (2) the basics of quantum computing. In each case, cognitive task analysis from both expert and student perspectives was either carried out directly or built upon from the results of prior investigations. I discuss the results of implementations of these learning tools in authentic classroom environments, which involves both online and in-person administrations and multiple instructors. In each case, student performance after engaging with the learning tools increased noticeably, and dramatically for some difficult concepts. [The dissertation citations contained here are published with the permission of ProQuest LLC. Further reproduction is prohibited without permission. Copies of dissertations may be obtained by Telephone (800) 1-800-521-0600. Web page: http://www.proquest.com/en-US/products/dissertations/individuals.shtml.]

Over decades anaerobic digestion (AD) has been successfully established as technology to treat organic wastes. The perspective of turning organic wastes into biogas, a source of renewable energy and profit, through a low-cost process, has gained increased interest around this technology. The AD as biological process is strongly dependent on the environmental conditions such as temperature, pH, presence of inhibitors, and nutrients content, that, in particularly unfavourable situations, can attribute to undesirable drop in performance and even for detrimental failures. Therefore, several strategies such as feedstock pre-treatment and process optimization have been developed to secure AD performance. However, these approaches do not remove inhibitor from the process, which may result in accumulation of the inhibitor and further destabilization of the AD system. From this point of view, it is beneficial to develop methods that remove and/or reduce the mobility and bioavailability the inhibitor within the digestion process. In this study, char derived from lignocellulosic wastes was added to anaerobic digestion to elucidate their roles on process performance and efficiency. The results of this analysis suggest that changes of AD performance (i.e. biogas production and intermediates concentrations) are strongly correlated to biochemical characteristics of char materials. Furthermore, the impact of changes in inhibitor concentrations on the digester's performance was observed, and found that inhibition tolerance of AD was improved by the presence of selected char materials. However, contrary to the initial hypothesis based on previous studies in related hydrochar application, the addition of RH hydrochar led to reduction in biogas production but an enhanced methane content in biogas was observed. Except from batch AD tests, a larger-scale AD system (semi-continuous stirring tank reactor) was used to determine the possibility of biochar on enhancing digestate structure and reduce the sensitivities towards environmental factors such as pH and other inhibitors. The further outcome from scale-up study revealed that biochar had improved ammonium tolerance (1.59 mg NH4-N/kg) by recovering 12-16.4% biogas production. This study also investigated effects of iron impregnated-thermal activation on physiochemical properties of the resultant biochar. Effects of these modified biochars on gas production and volatile fatty acids concentration was also measured and significant changes was observed. Overall, char material had a positive effect on AD system stability and promoted bio-methanation during the long run operations.

The present work numerically investigates the 3D flow structures around a canonical high-rise building immersed in an atmospheric boundary layer at different oncoming wind angles, using wall-resolved large eddy simulations. The switching between two vortex shedding modes is explored, and the influence of the atmospheric boundary layer on suppressing symmetric vortex shedding is identified. It is shown that the antisymmetric vortex shedding mode is prevalent in the near wake behind the building, with strong coherence between the periodic fluctuations of the building side force and the antisymmetric vortex shedding mode demonstrated. Exploiting this idea, active flow control strategies are designed to alleviate the aerodynamic side-force fluctuations. Two feedback control strategies are then developed to attenuate the building's unsteady loading when the oncoming flow is normal to the wider side of the building, using pressure sensing on just a single building wall. The sensor response to synthetic jet actuation along the two 'leading edges' of the building is characterised using system identification. Both the designed linear controller and the least mean square adaptive controller attenuate successfully the side-force fluctuations when implemented in simulations. The linear controller exhibits better performance, and its effect on the flow field is to delay the formation of dominant vortices and increase the extent of the recirculation region. Motivated by the effect of the downwash flow on the near wake of the building, an open-loop active control in the form of a synthetic jet located on the top surface is then considered. The aim is again to suppress side force fluctuations, this time exploring whether the simpler control setup can be effective across a wide range of oncoming wind angles. The behaviour of the synthetic jet and its effect on the building's unsteady side force, time-averaged flow fields and unsteady flow structures are investigated numerically. The synthetic jet actuation is found to reduce the side-force fluctuation of the building, enhance the downwash flow and successfully attenuate the antisymmetric vortex shedding. Finally, the possibility of using a robust feedback control strategy mitigating the unsteady loading of a high-rise building exposed to differing oncoming wind directions is explored. A reference linear open-loop transfer function is identified, and the differences between the reference open-loop system and open-loop transfer functions for different oncoming wind angles are assessed by v-gap. An H loop-shaping feedback controller is then designed and achieves the attenuation of the building's side-force fluctuations for differing oncoming wind angles. This work can provide a theoretical basis for the practical application of novel active control approaches to attenuating the side-force fluctuation of the high-rise building exposed to differing oncoming wind directions.

This thesis mainly deals with relationship between Taylor rules and exchange rate predictability for different inflation targeting(IT) countries. I first consider in Chapter 2 an exchange rate forecasting model in which movements in exchange rates are related to short-term nominal interest rates based on monetary policy rules. In this chapter, using forecasted data for the U.K. since instrument independence (1997- 2015), I find that a Taylor rule with expected inflation and expected output gap growth substantially improves short-run out-of-sample pound/dollar exchange rate predictability. This finding is firmly in line with intuition that under an IT regime where the central bank is largely forward looking, what matters for its monetary policy making is expected future variables. In Chapter 3, I propose the degree of central bank independence (CBI) as an explanation for the long-standing exchange rate disconnect puzzle in international macroeconomics, which is, the failure of macroeconomic fundamentals to predict exchange rates. To explore how changes in CBI influences the connection between exchange rates and macroeconomic variables, I first identify historical changes in CBI levels in the United Kingdom and compare them with structural changes in monetary responses to inflationary pressure. Considering CBI improvements in empirical models of the USD/GBP nominal exchange rate substantially improves their predictive power. In particular, over the whole sample period (October 1986- September 2008) the exchange rate disconnect puzzle remains dominant. However, when focusing on sub-periods following CBI increases, the Taylor rule model with UK forecasted variables helps predict exchange rates significantly better than the random walk model. These results support the hypothesis that CBI "reconnects" the exchange rate with macroeconomic variables through shifts of monetary policy regimes. Chapter 4 focuses on the exchange rate predictability in developing countries. I find that real exchange rate solely offers stronger evidence of out-of-sample predictability of nominal exchange rate than random walk and Taylor rule fundamentals model. Regarding the forecasts methodology, I also find that individual regression provides better forecasting accuracy than using the panel data regression. As different countries share same coefficient when using panel data estimation, the superior outcomes with individual regression implies that nominal exchange rate respond heterogeneously to initial movements in real exchange rate in developing countries. I use a robust set of out-of-sample statistics incorporating Diebold-Mariano Statistics, the Clark and West Statistics and Theil's U ratio. This finding is robust for one-month, six-month and twelve-month ahead forecasts.

Our digital world is awash with cookies, which are simple text files that keep website specific states on the web browser, such as auto-filled login fields. Although cookies are inherently harmless, third-party vendors use the tracking capability for commercial profits, e.g., cookie matching for audience targeting. This dissertation analyses browsing behaviour based on a large real user dataset collected by a browser extension developed, containing data of 2,537 users from 106 countries until August 2021 (from 10k+ installers). Then, providing solutions to inhibit third-party sharing/profiling and automated cookie protection tools. The first part studies the third-party ecosystem in different countries, revealing the impact of the type of first-party website sectors and the location of the user on the number of third parties in the wild. Results demonstrate that most users who are interested in a given site category are likely to encounter category-specific third parties, and around 65% of re-visited websites tend to offer more third parties to the same user profile. In terms of the user location, China is prone to a home-grown third-party ecosystem compared with the UK, due to China Great Firewall's access blockade of top third parties (i.e., Google, Facebook,etc.). To better understand the usage of cookies, I utilise the Cookiepedia database as the ground truth for a four-way classification (i.e., strictly necessary, performance, functionality and targeting/advertising cookies). The machine learning-driven framework achieves 94% F1 score and 1.5 ms latency, only 9.79% and 13.35% in the real-user dataset are identified as necessary and functional cookies. Briefly, most cookies are beneficial to the website rather than the user experience. After the preliminary analysis on the status quo, the dissertation proposes solutions to restrict cookie-based tracking for online behaviours from two aspects. One is a management assistant for multi-account containers for the reduction in third-party interconnectivity based on common third parties in browsing histories (i.e., "tangle factor"). Evidence shows that removing top third-party vendors does better than all ad blockers in decreasing interconnectedness. And uBlock origin is the best one among ad blockers, reducing the raw number of third parties by 60% and required containers by 40%. The other solution is the auto-processing of the GDPR minimal data option. Since May 26, 2018, the General Data Protection Regulation (GDPR) was promulgated in the EU to protect personal data without user approval. By the end of 2018, third-party cookies of UK users drop by over 10%. However, the consent fatigue and lack of an automatic consent setting mechanism resulted in the rebound of third-party cookies in 2019. Therefore, I build and deploy a browser extension to automatically assist users to protect user privacy in 85% of the websites with GDPR notices, reducing targeting/advertising cookies by 44.6%. Concisely, this dissertation mainly addresses the collection and classification of real-time browsing data in the wild, privacy risks of the third party interconnected tunnels and the lack of an automated GDPR-enforcing mechanism. And the field deployments increase the feasibility and usability, successfully hardening the protection against user privacy while browsing and paving the way for the automated global online privacy protection.

Organic electronics focuses on developing next-generation electronic devices, including organic photovoltaics, organic field-effect transistors and organic photodetectors, which rely heavily on the development of organic semiconductor materials, owing to their unique advantages like mechanical flexibility and stretchability, low processing temperature, low fabrication cost. This thesis focuses on the synthesis, characterization and application of novel organic semiconductor materials. Benzothiadiazole, as one of the most widely used electronwithdrawing units, has attracted great interest and extensive effort in the design of smallmolecular or polymerized semiconductor materials. Most of the approaches to tune the energetics of benzothiadiazole focuses on changing the heteroatom in the thiadiazole ring or functionalizing the 5 and/or 6 position of the benzo ring. Here in my work, we annulated an additional ring (2-(1,3-dithiol-2-ylidene)malonitrile, DTYM) in the 4,5-positions of the benzo ring for the first time, and coupled with indacenodithiophene core to get two small-molecular non-fullerene acceptors (BTSCN-IDT, FBTSCN-IDT). The photovoltaic performance of both acceptors was evaluated in organic photovoltaics when blended with donor polymer PM6, and the fluorinated acceptor FBTSCN-IDT delivered a comparatively better power conversion efficiency of 13.7% under dim light condition, which can be ascribed to the more red-shifted absorption and more coplanar molecular backbone. In addition, we annulated the same additional ring to the 5,6-positions of the benzothiadiazole to allow polymerization through 4,7-positions with four different donor units and received four polymer semiconductors (P1- P4). The effect of introducing DTYM group on benzothiadiazole was evaluated and found that the incorporation of DTYM group leads to a more twisted backbone but an enhanced dipole moment. As a result, all polymers presented n-type semiconductor behaviour in organic fieldeffect transistors, with the polymer P3 showing the highest saturation electron mobility up to 5.2 × 10-3 cm2 V-1 s-1 among them. However, most of the reported organic semiconductor materials have long conjugated molecular backbone which may induce chemical instability, low solvent solubility and high fabrication cost from the complicated synthesis. On this basis, two fully non-fused nonfullerene acceptors (BTIC-1, BTIC-2) were synthesized and applied in organic photovoltaics and photodetectors. Both acceptors share the same backbone but have the same alkoxy group on different positions to form varying number of intramolecular conformational locks to tune the molecular structure. Theoretical calculation and optical spectrum indicate that BTIC-2 presents a more planar conformation and thus a more red-shifted absorption up to 900 nm. As a result, BTIC-2 outperforms BTIC-1 in photodetector performance, with a low dark current of 2.4 × 10-7 A cm-2, leading to a high responsivity of 0.357 A W-1 and a specific detectivity of 1.7 × 1011 Jones at 828 nm under the reverse bias of -2 V. Lastly, we are also interested in developing novel fused-ring building blocks for constructing semiconductor materials. Two novel core molecules were designed and synthesized by bridging benzothiadiazole unit and adjacent thiophene units using bridging atom oxygen instead of the commonly used nitrogen atom, as oxygen is more electronegative than nitrogen which may lead to lower-lying energy levels and thus high open-circuit voltage.

This thesis describes the author's research contribution to the topics of SMPs and their applications. The contribution covers a period of 20 years, starting with the author's doctoral research and continuing with the research that the author subsequently carried out and directed, first as a professor in the Hong Kong polytechnic university and then in the city university of Hong Kong. The studies summarized here involve both experimental and numerical investigation to understand fundamental structure, mechanism, function, and applications of SMPs. The research is presented through the inclusion of 40 publications. All of them were published in scientific journals and are organized in four distinct groups. The first group includes publications on structure, mechanism, and function of SMPs. It includes studies to investigate the relationship between structure, mechanism, and function of SMPs, by designed and prepared different materials with shape memory effect. It also includes the studies of models for the SMEs for different SMPs. The second group contains publications of SMPs for textile applications. The studies are on the preparation of shape memory fiber, mainly focus on shape memory PU and natural wool hair. The sinning method, properties investigation, and theoretical explanation are included. The experiment for the development of shape memory textiles and their applications in the biomedical are also included. The third group consists of publications on SMB of natural biopolymers and their artificial version. The studies on the water sensitive shape memory model of natural biopolymer materials are included. The experiments for the artificial spider silks produced through biological and chemical methods, as well as their hierarchical structure, mechanical performance, SMB and other water responsive properties are also included. Another included topic is the summarizes the water sensitive shape memory phenomenon of other protein-based materials including keratin fibers and collagen skin. An overall summary precedes the publications, which first considers the research presented in each group separately, highlights the main contribution and points out how some of the idea generated have been taken up. A final section of the overall summary remarks the major findings of the research and shows the author's prospect of future research trends in the topic discussed in this thesis.

Perception is the foundation for intelligent robots to effectively explore environments and interact with users. Bio-inspired soft robots can exploit the compliance of their bodies, consequently demonstrating advantages in terms of safety during humanrobot interaction and operability in unstructured scenarios. However, this very feature dramatically increases the complexity of perception, which prevents soft robots from wide adoption in practical applications. This thesis aims at this challenge to implement a comprehensive study in coupling field simulation, sensor design and fabrication and learning-based perception algorithms. The major contributions of this thesis can be summarised as follows. First, coupling field simulation (CFS) was developed to integrate sensory systems into mechanical structures, which is critical for developing perceptive soft robots but neglected by most existing studies. The proposed simulation method was demonstrated through a 16-electrode capacitive sensor array deployed on a soft robot manipulator. The understanding of the sensor behaviours was built by simulating the sensor responses to a range of deformations using CFS. One of the most important applications of CFS is to generate annotated samples which can be used to train learning-based perception algorithms. The trained learning models could be transferred to practical scenarios without the need for tremendous annotated real-world data, thus reducing time and costs for data acquisition. Two case studies for applied force estimation and deformation classification were performed with annotated data generated by CFS to demonstrate its potential in developing learning-based perception methods. On soft robot proprioception, previous studies only achieved low proprioceptive geometry resolution (PGR), thus suffering from loss of geometric details (e.g., local deformation and surface information) and limited applicability (e.g., only applicable to prescribed simple deformations). This thesis proposed a high PGR soft robot proprioception system, which encapsulates an intrinsically stretchable capacitive e-skin (SCAS) and a capacitance-to-deformation transformer (C2DT), to endow full-geometry, millimetrelevel bodily awareness to soft robots. The SCAS has a redundant planar electrode layout that forms a sequence of capacitors sensitive to deformation across the entire soft robot. The C2DT based on transformer architecture can explore the dependency over the SCAS signals and recover deformation from the signals. The proposed proprioception framework synergistically combining the SCAS and the C2DT can achieve real-time (30 fps), high PGR (3,900) full-geometry deformation reconstruction with high accuracy (2.322 ± 0.687 mm CD error) under a range of complex deformations on a 20 × 20 × 200 mm soft manipulator. This high PGR was not demonstrated previously and is one or two orders of magnitude improvement over previous methods. Regarding tactile sensing (one of the most important exteroceptions), previous studies ignored the impact of deformation on the sensing data, making them a mismatch to practical scenarios where the geometric and tactile variations are coupled in the sensor signals. The thesis performed a preliminary exploration of tactile sensing with severe interference, which is inevitable in some practical applications. A simplified SCAS was proposed to achieve touch recognition with the interference induced by deformation while reducing the complexity of fabrication, deployment and wiring. The proposed method was validated on a pneumatic robotic platform. Contact location estimation was successfully achieved at low spatial resolution in various inflation conditions with 99.88% of classification accuracy. Moreover, deformation sensing with the interference induced by physical contact was also demonstrated by the coordinates estimation of 5 visual markers deployed on the soft robot platform. The C2DT was employed to perform marker tracking with the information of the first frame in a trajectory as prior knowledge and achieved 2.905±2.207 mm AD error, which shows the potential of the simplified SCAS to simultaneously detect internal and external stimuli. In summary, this thesis presented a framework for soft robot perception, which comprises stretchable capacitive e-skins to translate geometric and tactile variations to capacitance data and neural architectures to interpret the capacitance data to desired parameters. High PGR morphological reconstruction and tactile sensing with interference were demonstrated, paving the way towards the autonomy of soft robots.

While ornamental and recreational functions were traditionally the focus of urban parks in China, biodiversity conservation/enhancement has been increasingly seen as a prioritised goal. This promotes the use of the biodiversity-friendly approach in park design and the rise of a novel park style with a wild, less manicured appearance in Chinese cities. Such a park style contrasts with the traditional park style that has ornamental, manicured characteristics. Professionals state the benefits of the biodiversity-friendly approach, but the Chinese public's response to this new approach remains unclear. This research links the social and biodiversity values of Chinese urban parks by investigating how the public responds to the biodiversity-friendly approach to park design. More specifically, this research aims to 1) understand public reaction to biodiversity in urban parks, 2) investigate public perceptions of and preferences for the wild park style in contrast with the traditional style, and 3) examine preferences for park biodiversity and for the wild park style among the public's different sociocultural groups and between the public and ecology/landscape professionals. A mixed-method approach, comprising focus groups, conjoint analysis and semi-structured interviews, was applied to address the three objectives. The study was conducted in Hangzhou and Shanghai, two Chinese cities that have strong contextual and temporal legitimacy of the biodiversity-friendly park approach but where local residents' perceptions have been almost ignored. The key findings revealed that in general, the public in both Hangzhou and Shanghai demanded higher levels of biodiversity in parks. Residents of Hangzhou prioritised both biodiversity and facilities as the most important attraction factors of parks, whereas residents of Shanghai valued facilities the most and considered this factor more important than biodiversity. Residents of both cities generally showed a greater preference for the traditional park style over the wild style, yet preferences for the wild park style were enhanced when it demonstrated higher maintenance levels and contained more good facilities and semi-open vegetated areas. The study found the importance of ecology/landscape professional involvement and education for the appreciation of biodiversity in parks in Shanghai. Moreover, professionals preferred the wild park style and had a significantly higher preference for this style than the public in both cities. This research found that there was public demand for integrating higher biodiversity levels into park design but that this demand existed within a framework of good amenities and the traditional park style. This study's findings partially support existing policies and practices for the biodiversity-friendly development of urban parks in China. Public engagement and differences in residents' preferences compared with professionals should be incorporated into China's biodiversity-related landscape development. This study provides a reference for Chinese urban parks in search of win-wins of biodiversity improvement and social expectation.

The importance of self-complexity on leadership has been discussed by several theoretical studies. The present research draws upon Linville's (1985, 1987) social cognitive perspective of self-complexity, which is defined as a function of two things: the number of aspects that one uses to cognitively organize knowledge about the self, and the degree of relatedness of these aspects" (Linville, 1985, p. 97), to serve as one of the limited empirical studies that examine its associations with leadership behaviors. To assess self-complexity, I develop a measurement that focuses on capturing the structure of the self at the role level and adopt the pairwise comparison method to enrich our understanding of self-complexity. In addition, I also draw upon self-control and goal shielding theories to examine the psychological factors such as time and centrality in self-complexity. Regarding leadership behaviors, I studied two often-seen leadership behaviors in the current research. For positive leadership behavior, I study servant leadership, which refers to "an (1) other-oriented approach to leadership (2) manifested through one-on-one prioritizing of follower individual needs and interests, (3) and outward reorienting of their concern for self towards concern for others within the organization and larger community" (Eva, Robin, Sendjaya, Van Dierendonck, & Liden, 2019, p.114). For the negative leadership behavior, I study abusive supervision, which is defined as "subordinates' perceptions of the extent to which supervisors engage in the sustained display of hostile verbal and nonverbal behaviors, excluding physical contact" (Tepper, 2000, p.178). To study the associations between self-complexity and leadership behaviors, I employ two different perspectives: Self-control and cognitive theories. From the self-control perspective, I adopt self-control and conservation of resource theories to examine the indirect associations between self-complexity and leadership behaviors via ego depletion, which is defined as "a state in which the self does not have all the resources it has normally" (Baumeister & Vohs, 2007, p.116). From the cognitive perspective, I employ cognitive theories to examine the indirect associations between self-complexity and leadership behaviors via cognitive flexibility, which is defined as "the ability of individuals to restructure knowledge in multiple different ways depending on changing situational demands" (Gino & Ariely, 2012, p.446). Four studies were conducted in the current research. The results in Study 1 (U.K. sample, N = 179) showed that the number of self-aspects and the degree of overlap among self-aspects were negatively associated with ego depletion and were positively associated with cognitive flexibility. Moreover, ego-depleted leaders were found to be more likely to behave abusively toward their followers, whereas cognitively flexible leaders were found to be less likely to abuse their followers. Regarding the psychological factors, results in Study 1 showed that the relationship between the interaction of a high number of self-aspects and a high degree of overlap among self-aspects and ego depletion was weaker when the time was short rather than high. The results in Study 2 (Taiwan sample, N = 332) also showed that the degree of overlap among self-aspects was negatively associated with ego depletion and positively associated with cognitive flexibility. Nevertheless, the number of self-aspects was not found to be negatively associated with ego depletion nor positively associated with cognitive flexibility. Study 2 also showed similar findings regarding leadership behaviors; ego depletion was positively associated with abusive supervision, and cognitive flexibility was negatively associated with abusive supervision. Study 3 specifically focused on testing the differences between the traditional pen and paper trait sorting task (N = 98) and the newly introduced method of measuring self-complexity (N = 110) and on examining the positive leadership behavior (i.e., servant leadership) in the current research model. The results in Study 3 showed that both methods had similar findings to Studies 1 and 2. However, the newly introduced method that measures the structure of the self at the role level seems to be a better approach than Linville's free-response format in predicting ego depletion and cognitive flexibility. In addition, ego-depleted leaders were not found to be less likely to display servant leadership in the workplace, and cognitively flexible leaders were found to be more likely to display servant leadership behaviors in the workplace. The results in Study 4 (N = 144 leaders and N = 450 followers) showed that the number of self-aspects and the degree of overlap among self-aspects were negatively associated with ego depletion and were positively associated with cognitive flexibility. Moreover, ego-depleted leaders were more likely to be perceived as abusive supervisors, and cognitively flexible leaders were less likely to be perceived as abusive supervisors and more likely to be perceived as servant leadership by their followers.

This thesis asks how East Asian identities are performed in contemporary British theatre. The research is conducted through analyses of, and discourses surrounding performances that took place between 2016 and 2019. It explores the strategies adopted by East Asian artists in Britain (EAB) and their performance practices. What does it mean to be an East Asian artist in Britain? How is East Asian identity shaped by historical, cultural, material conditions, and what kind of continuity and subversion does it undergo through performance practice? This thesis is divided into four chapters. Chapter 1 explores the historical context of this research in relation to representation of China and East Asia in the UK. The following three chapters examine contemporary performance productions, each focusing on a main theme: 'universality', 'authenticity', and 'cultural currency'/'consumption'. Chapter 2 on universality examines performances of or by East Asian people aiming/claiming to represent universal values. The cases studied in this chapter are In the Depths of Dead Love (the Print Room, 2017) and Tamburlaine (Yellow Earth Theatre, 2017). Chapter 3 considers performance and representation of East Asian identities situated within a specific time-space, adopting theories of 'languaging' to explore how East Asian artists tune their own identities. The case studies in this chapter are Snow in Midsummer (the RSC, 2017) and Forgotten (Yellow Earth Theatre and Moongate Productions, 2018). Chapter 4 shifts the focus to solo auto/biographical performances framed through a concept I call 'consumption of identity'. I examine my own performance practice from 2016 to 2019 in this chapter, and consider the strategies individuals employ to simultaneously challenge and make use of structural stereotypes. I argue that under the broad, contested, category of 'East Asian', East Asian artists from diverse cultural backgrounds adopt different artistic strategies at different stages in sometimes self-contradictory ways. On the whole, performance of East Asian identity is as fluid as the identity itself, and is at the same time a restriction and a resource in performance practice.

Sampling is an important topic in the area of computational statistics. Being able to draw samples from a designated distribution allows one to numerically compute various statistics without the need to solve for solutions analytically. A popular branch of the sampling method generates samples by evolving a stationary Markov chain that admits the target distribution as its stationary distribution. The problem, however, is that one does not have a universal criterion to assess whether the chain is stationary. On the other hand, exact simulation methods, being the focus of this thesis, always produce samples that precisely follow the target distribution. We first begin with the path-space rejection sampling for the exact simulation of diffusion bridges and show how this rejection scheme can be further set up into an exact simulation method for sampling product densities. We provide guidance on how to tune the algorithm parameters in order to attain a near-optimal performance and introduce the construction of an importance sampler/particle filter based on the same theoretical result for better efficiency. Finally, we show a variant of the sampler that deals with linear constraints which render most of the target distributions intractable. Two application studies are conducted in the end to demonstrate the effectiveness of the algorithm.

Zinc and natural ligands of the aryl hydrocarbon receptor (AHR) are dietary components that regulate intestinal epithelial barrier function. Although zinc has been well-studied in this context, its mechanism of action is inconclusive. Moreover, AHR has several reported functions in the gut but its function in the intestinal epithelial barrier function remains relatively unexplored. My hypothesis was that dietary zinc combined with an AHR agonist improves epithelial barrier function of the human intestine. To verify this hypothesis, I investigated the regulation of gene transcripts involved in zinc homeostasis by AHR activation and delineated the mechanisms by which zinc and AHR activation improve barrier function. AHR activation by 6-Formylindolo[3,2-b] carbazole (FICZ) was shown to regulate transcripts of zinc-regulatory genes, including metal transcriptional factor-1, metallothionein-1, SLC39A2, SLC39A4, SLC39A7 and SLC39A10, concurrently with changes in cytosolic zinc concentrations in differentiated Caco-2 cells. These results were confirmed in human ileum organoids and wild-type/AHR-Villin-KO mice organoids. Additional data showed that AHR activation by FICZ and zinc supplementation improved the barrier function of Caco-2 cells when challenged with dextran sodium sulphate (DSS) and hypoxia, and that FICZ and zinc maintain tight junctions (TJs) by preventing activation of the NF-κB pathway and inhibiting calpain activity. Furthermore, addition of the zinc chelator Tris(2-pyridylmethyl) amine (TPA) confirmed that AHR activation by FICZ increases TJ function in Caco-2 cells by mediating zinc homeostasis.

Language is one of the hallmarks of intelligence, demanding explanation in a theory of human cognition. However, language presents unique practical challenges for quantitative empirical research, making many linguistic theories difficult to test at naturalistic scales. Artificial neural network language models (LMs) provide a new tool for studying language with mathematical precision and control, as they exhibit remarkably sophisticated linguistic behaviors while being fully intervenable. While LMs differ from humans in many ways, the learning outcomes of these models can reveal the behaviors that may emerge through expressive statistical learning algorithms applied to linguistic input. In this thesis, I demonstrate this approach through three case studies using LMs to investigate open questions in language acquisition and comprehension. First, I use LMs to perform controlled manipulations of language learning, and find that syntactic generalizations depend more on a learner's inductive bias than on training data size. Second, I use LMs to explain systematic variation in scalar inferences by approximating human listeners' expectations over unspoken alternative sentences (e.g., "The bill was supported overwhelmingly" implies that the bill was not supported unanimously). Finally, I show that LMs and humans exhibit similar behaviors on a set of non-literal comprehension tasks which are hypothesized to require social reasoning (e.g., inferring a speaker's intended meaning from ironic statements). These findings suggest that certain aspects of linguistic knowledge could emerge through domain-general prediction mechanisms, while other aspects may require specific inductive biases and conceptual structures. [The dissertation citations contained here are published with the permission of ProQuest LLC. Further reproduction is prohibited without permission. Copies of dissertations may be obtained by Telephone (800) 1-800-521-0600. Web page: http://www.proquest.com/en-US/products/dissertations/individuals.shtml.]

The first article is a literature review that analyzed 32 articles on computer-assisted L2 Chinese learning in higher education from 2005-2021. Findings revealed current trends, such as mobile-assisted language learning (MALL) and the use of Web 2.0 technologies, and the importance of speaking as the most researched language skill. Most studies employed mixed methods and showed the positive effects of computer-assisted learning on L2 Chinese skills. Challenges identified included limited access to technology, technical difficulties, lack of time and support, and insufficient online/blended pedagogies and competencies. The second article used Ajzen's Theory of Planned Behavior to examine L2 Chinese instructors' intentions to teach through blended modalities and their perceptions of its challenges and benefits. We used a mixed survey methodology to collect data from 51 North American instructors. Findings showed that instructors' attitudes, subjective norms, and perceived behavioral control influenced their intentions to adopt blended teaching. They identified some concerns about blended learning drawbacks, mixed perceptions of stakeholder approval and the need for institutional support, training, and resources. Despite challenges, they also provided valuable recommendations for effective blended learning strategies. The study highlights the importance of institutional support, technological expertise, and well-designed course structures for successful blended L2 Chinese courses, suggesting increased institutional support, comprehensive teacher training, professional communities, and further research on blended learning's impact on L2 Chinese education. The third article is a qualitative study that explored L2 Chinese language instructors' experiences with online and blended teaching methods in U.S. higher education after the COVID-19 pandemic. The findings from semi-structured interviews with seven participants revealed a shift towards student-centered approaches, prioritizing student needs, interests, and mental well-being. Instructors are adapting to new classroom dynamics, emphasizing professional development and effective use of technology. The debate over handwriting versus typing Chinese characters calls for further research on efficient teaching methods in online and blended learning. The study highlights the importance of adapting and refining pedagogical practices to serve students in a post-pandemic world better. [The dissertation citations contained here are published with the permission of ProQuest LLC. Further reproduction is prohibited without permission. Copies of dissertations may be obtained by Telephone (800) 1-800-521-0600. Web page: http://www.proquest.com/en-US/products/dissertations/individuals.shtml.]

The research aimed to provide a complete understanding of characteristics, facilitators, and challenges in implementing professional learning communities in multi-age schools. This study utilized a mixed-methods multiple case study design. Participants in this study were teachers from two private K-8 multi-age schools. The characteristics of professional learning communities and their influence on school quality and student performance were identified. Results confirmed that experiences in professional learning communities increased teacher involvement in all six domains, especially in collective responsibility, reflective dialogue, and peer collaboration. Results indicated that professional learning communities explained significant variance in teacher-perceived school quality measured by academic expectations, communication, and engagement. The sustainability of a successful professional learning community is vital to school quality. The insight gained from this study revealed several factors that enhance the development of professional learning communities in multi-age schools. The first theme, school structural conditions, included subcategories such as time, space, technology and instructional materials, communication systems, district support, and school size as facilitators of professional learning communities in multi-age schools. The second theme, school human and social resources included parental involvement and support, openness to innovation/improvement, professional development, and supportive and caring relationships between students and teachers. The third theme was shared and supportive leadership which included the roles of principal and teacher leader role. Results from this study demonstrated that lack of time, technology, supplies, instructional materials, and parent support were the top challenges in teaching and implementing professional learning communities in multi-age schools. This study offered suggestions for school leaders looking to make practical and authentic changes to improve professional development during school transformation. [The dissertation citations contained here are published with the permission of ProQuest LLC. Further reproduction is prohibited without permission. Copies of dissertations may be obtained by Telephone (800) 1-800-521-0600. Web page: http://www.proquest.com/en-US/products/dissertations/individuals.shtml.]

Giving machines the ability to precisely perceive and understand the 3D visual world is the fundamental step to allow them to interact competently within our physical world. However, the research on large-scale 3D scene understanding and perception is still in its infancy, due to the complex geometrical structure of 3D shapes and limited high-quality data resources. Among various 3D representations, point clouds have attracted increasing attention due to its flexibility, compactness, and the characteristic of being close to raw sensory data. Despite this, the semantic understanding of large-scale 3D point clouds remains a challenge due to its orderless, unstructured, and non-uniform proper-ties. To this end, this thesis makes three core contributions from a high-quality urban-scale dataset, and then fully-supervised semantic understanding to weakly supervised label efficient learning of large-scale 3D point clouds. The main contributions of this thesis are three-fold. In Chapter 3, we start by building an urban-scale photogrammetric point cloud dataset with nearly three billion richly annotated points. We also identify a number of open and unique challenges faced by urban-scale 3D scene understanding and conduct comprehensive experimental analysis to address these challenges. This work points out the limitations of existing algorithms and provides several insightful conclusions for understanding large-scale 3D point clouds in urban environments. With high-quality large-scale 3D datasets, we further investigate the research problem of efficient semantic segmentation of large-scale 3D point clouds in Chapter 4. We begin by analyzing the strengths and weaknesses of existing downsampling strategies and find that random sampling is a suitable component for efficient learning on large-scale point clouds. Additionally, we propose a local feature aggregation module to increase the receptive field hierarchically and preserve important features. We then build an efficient, lightweight neural architecture called RandLA-Net that can directly infer per-point semantics for large-scale point clouds with millions of points that span hundreds of meters. In Chapter 5, we take a step further to investigate the label-efficient learning of large-scale 3D point clouds. That is, achieving high-quality semantic segmentation with limited annotations. We first examine key issues in weakly supervised learning of 3D point clouds, including different weak supervision schemes and the critical point of weak annotations. Through a pilot study, we find that dense 3D annotations are actually redundant and unnecessary. Motivated by this, we propose a novel weak supervision framework that implicitly augments the total amount of available supervision signals by leveraging the semantic homogeneity between neighboring points. This is achieved using a point neighborhood query to allow the sparse training signals to be back-propagated to a wider context. All of the algorithms and datasets presented in this thesis have been open-source on GitHub to facilitate future research. The RandLA-Net algorithm was recognized as one of the most influential papers at the CVPR 2020 conference and has been integrated into code libraries such as Open3D and TorchPoints3D. The SensatUrban dataset was used as the platform for two Urban3D challenges, contributing to the advancement of semantic understanding of city-scale point clouds. Overall, this thesis presents a high-quality dataset and two novel data-driven algorithms, aiming to achieve efficient, scalable, and effective learning-based semantic understanding of large-scale 3D point clouds, eventually improving the real-time 3D perception capacity of intelligent machines in practice.

Climate change and fast urbanization are increasing the likelihood of compound hazards - events where multiple drivers and/or hazards interact with multiplicatively destructive environmental and socio-economic consequences. This includes increases in the frequency of not only concurrent natural extremes (heatwaves and droughts, storm surges and extreme rainfalls, etc.), but also collisions between natural and manmade disasters (air pollution, infectious disease transmission, trade wars, etc.) particularly in the post-pandemic world. Entanglement of different hazardous factors increases the complexity of impact accounting and risk management and requires an integrated solution to tackle the vulnerabilities of human societies towards compound risks. However, most of the research in disaster analysis investigates one hazard at a time. Only a few emerging perspectives have noticed or warned the potential of compound hazards, but they are still far from capacity building for the compound resilience to future crises. This PhD thesis presents a full set of methodology to systematically assess the economic impacts from single to compound hazards. The concept of 'disaster footprint' is used here to capture the direct and indirect impacts rippling through the economic supply chain during a single or compound disaster event. A four-stage research framework is proposed. It starts from the direct disaster footprint assessment, which links physical characteristics of hazards with property damage or health impairment by simulating hazard-specific exposure-damage functions. The direct footprint is then fed into an input-output-based (IO-based) hybrid economic model to calculate the indirect disaster footprint that propagates through intersectoral and interregional connections to wider economic systems. The improved IO-based disaster footprint model is built here for single hazard analysis, with innovations regarding inventory adjustment and cross-regional substitutability. Third, within the same disaster footprint framework, the economic interplays between diverse types of hazards are synthesized into the impact assessment, and thereby a Compound Hazard Economic Footprint Assessment (CHEFA) model is developed for compound events. Finally, favourable response and recovery plans, which are aimed to mitigate the total disaster footprint, are suggested by comparing the modelling results under wide ranging scenarios and identifying crucial influencing factors through sensitivity analysis. A major contribution of this thesis is that it takes the first step in the field of disaster analysis to integrate multiple hazardous factors within a macro-economic impact assessment framework that accounts for both direct and indirect disaster footprint into sectoral and regional details. The proposed modelling framework is first applied to three types of hazards (i.e., heat stress, air pollution and climate extremes) on the provincial and national scales in China to demonstrate its flexibility for a wide range of disaster risks. The total economic costs of heat stress, air pollution and climate extreme events in China have increased from US$207.9 billion (1.79% of GDP) in 2015 to US$317.1 billion (2.16% of GDP) in 2020. Despite the decreasing economic costs of air pollution and climate extreme events, the economic costs from heat-related health impacts have continued the concerning growing trend. Among the three types of hazards, the economic costs of heat stress were the biggest and accounted for over 70% of the total costs. Heat stress affects the economy mainly by reducing labour productivity. For each unit of direct costs, heat stress was also inclined to cause more indirect supply chain costs than air pollution and climate extremes. Most of the heat-induced direct costs occurred outdoors in the agriculture and construction sectors, while most of the heat-induced indirect costs happened indoors in the manufacture and service sectors. At the regional level, hotspot provinces with prominent economic risks from these hazards have been identified for China. Southern provinces were more economically vulnerable to heat stress than northern provinces, while northern provinces tended to suffer larger economic costs from air pollution than southern provinces. By contrast, the economic impacts of climate extreme events were more spatially distributed in China than the other two types of hazards. Location-specific economic impacts of climate change require location-specific responses, including enhancing inter-departmental cooperation, strengthening climate emergency preparedness, supporting scientific research, raising public awareness, and promoting climate change mitigation and adaptation. Economic implications of climate change are also evaluated with a focus on future flood risks in six developing countries (i.e., Brazil, China, India, Egypt, Ethiopia and Ghana) around the end of 21st century (2086-2115). A physical model cascade of climate-hydrological-flood models is linked with the disaster footprint economic model through a set of country and sector specific depth-damage functions. The total (direct and indirect) economic losses of fluvial flooding are projected for each country, with or without socio-economic development, under a range of warming levels from < 1.5°C to 4°C. As a share of national GDP, Egypt suffers the largest flood-induced losses under both climate change (CC) and climate change plus socio-economic development (CC+SE) experiments, reaching 2.3% and 3.0% of GDP under 4°C warming. Climate change acts as a driving factor that increases the flood losses in all countries, but the effect of socio-economic development differs among the countries and warming levels. For Ethiopia and China, future flood losses as a proportion of GDP under different warmings decline from the baseline levels when socio-economic development is modelled, suggesting a more resilient economic growth that helps reduce future flood risks. However, for Brazil, Ghana, and India, while losses as a proportion of GDP initially decline at lower warming levels, increases are seen from 2.5°C or 3°C warming onwards, suggesting a tipping point where increasing flood risk outweighs any relative benefits of socio-economic development. These results highlight the importance of including socio-economic development when estimating future flood losses, essential to provide a more comprehensive picture of potential losses that will be important for decision makers. With the development of the CHEFA model, the economic interaction between concurrent hazardous factors comes into analysis. A hypothetical perfect storm consisting of floods, pandemic control, and trade restrictions (as a proxy for deglobalization) is assumed to test the applicability and robustness of the model. The model also considers simultaneously cross-regional substitution and production specialization, which can influence the resilience of the economy to multiple shocks. Scenarios are first designed to investigate economic impacts when a flood and a pandemic lockdown collide and how these are affected by the timing, duration, intensity/strictness of each event. The results reveal that a global pandemic control aggravates the flood impacts by hampering the post-flood capital reconstruction, but a flood exacerbates the pandemic impacts only when the flood damage is large enough to exceed the stimulus effect of the flood-related reconstruction. Generally, an immediate, stricter but shorter pandemic control policy would help to reduce the economic costs inflicted by a perfect storm. The study then examines how export restrictions and retaliatory countermeasures during the pandemic and floods influence the economic consequences and recovery, especially when there is specialization of production of key sectors. It finds that the trade restriction of a region to 'protect' its product that can be substituted by the same product made elsewhere, while hampering the global recovery, may alleviate the region's own loss during the compound disasters if the increasing domestic demand exceeds the negative impacts of falling exports. By comparison, the trade restriction on a non-substitutable product has greater negative impacts on the global recovery, which ultimately propagates backward to the region through the supply chain and exacerbates its own loss. The results also indicate that the potential retaliation from another region and sector would further deteriorate the global recovery and make everyone lose, with the region which initiates the trade war losing even more when the retaliatory restriction is also imposed on a non-substitutable product. Therefore, regional or global cooperation is needed to address the spillover effects of such compound events, especially in the context of the risks from deglobalization.

In recent years, the Internet of Things (IoT) has grown to be an essential part of the development of tourism, with an increasing number of IoT solutions being applied in tourism-related industries such as hospitality and transportation. Despite increasing academic investigations of the role of IoT in the tourism industry from the traveller's perspective, most studies have focused on the traveller's behaviour, experiences, and satisfaction. It is not clear how IoT affects traveller subjective well-being which is a quintessential element of individual decision-making in particular tourism markets. Therefore, the purpose of this study was to propose a theoretically integrated approach to understanding the influences of IoT on traveller subjective well-being in the context of city breaks. In doing so, the present study explored the factors and the extent to which IoT affect city break traveller subjective well-being on the basis of task-technology fit theory, expectation confirmation model, and Hofstede's culture theory. A quantitative method was adopted in this study. Empirical data were collected from 2077 city break travellers, 1008 in the UK and 1069 in China. An integrated model was tested using the structural equation modelling (SEM) approach. The findings of the study revealed that the main factors that determine city break traveller subjective well-being towards the use of IoT are a combination of technological, emotional, personal, and contextual attributes. These findings contribute to the theoretical development of innovative technologies and traveller subjective well-being in general as well as enrich the literature regarding city break tourism. In addition, the study discusses some practical implications for the marketing of city break tourism with the use of IoT.

Content-based Image Retrieval (CBIR) is the task of searching for the most similar images to the query content from an extensive image database. Most existing feature extraction methods and attention mechanisms for CBIR tasks are query non-sensitive, ignoring the specifics of the query pattern, which may lead to focusing on irrelevant regions to the query content. In this thesis, we explore query sensitive attention mechanisms for CBIR task, which involves query feature information in the feature extraction procedure of the candidate image. Firstly, we propose the Conditional Attention Network (CANet). CANet takes the query image and a candidate image as input, resulting in a co-attention map of the candidate image under the condition of the query content. The generated co-attention map could correctly highlight the target object and improve image retrieval performance when embedded into a convolution neural network (CNN) based feature extraction pipeline. Secondly, another more efficient co-attention method is proposed based on local feature selection and clustering over candidate local features. Using local feature selection and clustering dramatically reduces the computation costs caused by the query sensitivity but still leads to accurate co-attention maps even under challenging situations. The proposed clustering-based co-attention method leads to new state-of-the-art performance on several benchmark datasets. Lastly, we explore using clustered expressive local features to perform many-to-many local feature matching for CBIR. We show that the proposed local feature matching method implicitly generates co-attention-like local matching maps. In addition, a trainable binary encoding layer is applied for network fine-tuning, enabling the model to generate compact binary codes with slight performance degradation and greatly reducing computation costs. In summary, we demonstrate that the query information could play an important role in feature extraction for the CBIR task. With a simple design, co-attention could be practical and effective even for large-scale image retrieval tasks.

Sexual objectification is the experience of being treated as a body existing for the use and pleasure of others. When sexually objectified, individuals may experience self-objectification and other body image concerns. It is currently unclear whether existing measures of self-objectification, which are primarily designed for women, adequately capture self-objectification in cisgender heterosexual men. In addition, current measures of experiences of sexually objectifying media may not fully capture the experiences of cisgender heterosexual women and men. Six studies were conducted in this PhD to address these gaps in the literature. Studies 1 and 2 examined the psychometric properties of the Self-Objectification Questionnaire (SOQ; Noll & Fredrickson, 1998), the Objectified Body Consciousness Body Surveillance Scale (OBC-Surv; McKinley & Hyde, 1996), and the Self-Objectification Beliefs and Behaviors Scale (SOBBS; Lindner & Dunn, 2017) in cisgender heterosexual women and cisgender heterosexual men using a longitudinal study design. Studies 3a, 3b, 4, and 5 developed and evaluated two novel measures of sexually objectifying media experiences for age-representative samples of cisgender heterosexual women (Women-SOMS) and men (Men-SOMS). Studies 3a and 3b generated the initial item pools for the Women-SOMS and the Men-SOMS based on a literature review and two online surveys. Study 4 conducted Exploratory Factor Analysis for both measures. Study 5 conducted Confirmatory Factor Analysis and examined the validity and reliability of the Women-SOMS and the Men-SOMS. Studies 1 and 2 indicated that the SOBBS was the most robust measure of self-objectification for cisgender heterosexual women and men. Studies 3a, 3b, 4, and 5 demonstrated that the Women-SOMS and Men-SOMS are generally psychometrical sounds for measuring experiences of sexual objectification in the media. Future research should continue to investigate the validity of the SOBBS, Women-SOMS and Men-SOMS in women and men.

Machine Learning is widely applied in industrial applications such as product design and recommendations, sales and churn prediction among many others. This thesis proposes scalable methods geared towards industrial application where we address problems related to dimensionality, uncertainty quantification and inference. First, we inferring features from images of products that are pertinent to high sales. We propose an additional regularization objective inspired by kernel two-sample testing to separate high-sellers from low-sellers in latent space for generative models. Experiments demonstrate that we can achieve linear separation in latent space and consequently infer important features using lasso methods. Next, we develop a non-parametric test for causal association. We propose the backdoor-Hilbert Schmidt Independence Criterion with importance weights that encode the interventional distribution. Experiments demonstrate that the proposed test is calibrated and has good power. We then propose Kernel Fried Tensor (KFT), which is a tensor factorization model augmented with kernelized side information, designed to handle high-dimensional forecasting. To improve the utility KFT, we derive a variational inference scheme to provide uncertainty quantification around our estimates. Experiments demonstrate competitive forecasting results and calibrated uncertainty quantification. Further, we consider right-censored data where existing survival models do not have a tractable likelihood and evaluate on non-proper-scores. We propose survival monotonic neural networks (SuMo-net) to train and evaluate on the right-censored likelihood, a proper score. Experiments demonstrate that SuMo-net achieves improved likelihood scores with lower inference times. As many of our proposed methods rely on kernel matrix vector multiplication (KMVM), they are bottlenecked by the O(n²) memory and complexity. We propose Faster Fast and Free Memory Method (F³M) and achieve O(n) memory and complexity for tall (n 10⁹) and skinny data (D ≤ 7). In summary, the methods proposed in this thesis provide a collection of scalable methods for various use cases often encountered in industrial settings and contributes both through novel methodology and statistical characterization of the tasks, which hopefully inspires further research in the machine learning community.

The intergenerational transmission of education examines to what extent has one generation's attainment determined the next generation's attainment. It is deemed an important indicator of social fluidity and inequality. This thesis explores the demographic pathways of intergenerational transmission, focusing on fertility and assortative mating. It includes four self-contained essays, two on fertility and two on assortative mating. Chapter One brought the previous individual-level inquiries on the relationship between sibship size and educational attainment to the societal level. Using the case of China, findings show that fertility decline during its One Child Policy era brought modest improvement in the average educational attainment of the country. However, differential fertility rates by rural/urban residency have a marked impact on rural/urban educational inequality. Chapter Two examined the role of fertility in three-generational transmission. It proposed a new moderator for the process: the fertility rates of the grandparents. Findings using data from the Panel Studies of Income Dynamics (PSID) show that having multiple children (i.e., multiple members in the parental generation) amplifies grandparents' influence on grandchildren's educational attainment. Chapter Three examined the role of parents in adult children's marital sorting. It extends previous findings on first and new marriages from a life course perspective. Results using PSID data show divergent trends by gender. Parents' influence becomes less important for males but more vital for females over the life course. Chapter Four examined the role of education homogamy in intergenerational reproduction. It brought the previous one-sided analysis, which focused on one parent only, to the couple level and proposed an approach that is the first to consider and distinguish two sources-resource combination and family milieu-that generate the difference between homogamy and heterogamy. Findings using PSID data demonstrate that these two sources affect couples' probability of having children and their children's educational attainment differently, and their influences change over cohorts.

Humans navigate in their environment by learning a mental model of the world through passive observation and active interaction. Their world model allows them to anticipate what might happen next and act accordingly with respect to an underlying objective. Such world models hold strong promises for planning in complex environments like in autonomous driving. A human driver, or a self-driving system, perceives their surroundings with their eyes or their cameras. They infer an internal representation of the world which should: (i) have spatial memory (e.g. occlusions), (ii) fill partially observable or noisy inputs (e.g. when blinded by sunlight), and (iii) be able to reason about unobservable events probabilistically (e.g. predict different possible futures). They are embodied intelligent agents that can predict, plan, and act in the physical world through their world model. In this thesis we present a general framework to train a world model and a policy, parameterised by deep neural networks, from camera observations and expert demonstrations. We leverage important computer vision concepts such as geometry, semantics, and motion to scale world models to complex urban driving scenes. In our framework, we derive the probabilistic model of this active inference setting where the goal is to infer the latent dynamics that explain the observations and actions of the active agent. We optimise the lower bound of the log evidence by ensuring the model predicts accurate reconstructions as well as plausible actions and transitions. First, we propose a model that predicts important quantities in computer vision: depth, semantic segmentation, and optical flow. We then use 3D geometry as an inductive bias to operate in the bird's eye view space. We present for the first time a model that can predict probabilistic future trajectories of dynamic agents in bird's-eye view from 360◦ surround monocular cameras only. Finally, we demonstrate the benefits of learning a world model in closed-loop driving. Our model can jointly predict static scene, dynamic scene, and ego-behaviour in an urban driving environment. We show that the learned world model and driving policy can generate predictions of over 1 hour (2000x longer than the training sequence size).

Levitated optomechanics opens the door for many quantum experiments and sensing applications with the advantage of minimising the dissipation to the environment. It provides a unique research platform to control and cool the motion of micro/nano-size objects to the quantum regime, pushing the mass limit of exploring macroscopic quantum phenomena. Apart from the centre-of-mass motion of levitated nanoparticles, control and cooling of rotational degrees of freedom are of significant research interest, which explores another avenue for studying fundamental physics and developing quantum technologies. This thesis focuses on ro-translational optomechanics with optically levitated nanoparticles in vacuum. A 1550 nm counter-propagation standing-wave optical trap experimental platform has been established for studying levitated optomechanics in high vacuum. Direct loading, optical trapping and corresponding detection techniques for high-efficiency measuring ro-translational degrees of freedom of levitated nanoparticles have been built. A clean, vacuum-compatible method based on laser-induced acoustic desorption (LIAD) is developed for directly loading nanoparticles into the optical trap, enabling us to load tailored silicon nanorods from the substrate with very high efficiency. The translational and rotational dynamics of levitated nanoparticles in vacuum have been systematically studied. Besides driving the levitated silicon nanorotor by circularly polarised light, it has been driven to full rotations at tens of MHz in several millibars in the transverse plane by transferring transverse orbit angular momentum, achieving five orders of magnitude higher applied torque compared to other reported nanorotors with the same optical power. Furthermore, the ro-translational motion of levitated nanoparticles can be cooled by implementing feedback cooling schemes, such as parametric feedback for centre-of-mass and polarization feedback for librational motion. These paves way for next-step tests of rotational macroscopic quantum superposition experiments, such as orientational quantum revivals.

The Chinese population is ageing rapidly, and older Chinese adults (aged 45 and over) have spent a large proportion of their lives exposed to historically high levels of air pollution (AP). Previous studies suggest that there is a strong association between AP and individual health outcomes, but most of them have relied on cross-sectional measures of AP exposure and health, or both. These limit our understanding of how short-term, long-term, and cumulative exposures lead to changes in health outcomes. This PhD thesis addresses these gaps by investigating the longitudinal relationship between long-term and short-term exposure to AP and the health of older adults in China. The thesis uses two kinds of survey data and two kinds of AP data. The Chinese Longitudinal Healthy Longevity Survey (CLHLS) 2011 and 2014 is linked with monitoring station AP data (used in chapter 4), and the China Health and Retirement Longitudinal Survey (CHARLS) 2011, 2013 and 2015, is linked with PM₂.₅ data that are retrieved from satellite data via remote sensing technology (used in chapters 5 and 6). The main finding of the first empirical chapter using the CLHLS is that increased frailty incidence is associated with higher long-term exposure to AP rather than short-term fluctuations. The second empirical chapter using the CHARLS shows that exposure to PM₂.₅ over 15 years is associated with higher multimorbidity accumulation, and higher levels of PM₂.₅ are associated with a higher likelihood of membership to both respiratory and cardio-metabolic disease classes. The results of the third empirical chapter indicate that long-term exposure to PM₂.₅ is associated with poorer cognitive function, but different measures of long-term AP exposure are associated with different levels of cognitive decline. Overall, the three chapters indicate that long-term AP exposure has a negative association with elderly health, but that there may be some individual and contextual differences which are being conflated with these associations, and which deserve further exploration. This thesis, therefore, makes several substantive and methodological contributions. First, I create a unique longitudinal dataset by linking historical AP data to representative longitudinal ageing surveys at the city level. Second, I investigate the associations of short-term and long-term exposure with health outcomes. Third, health outcomes of older adults are measured longitudinally using three comprehensive indicators that can reflect the complex relationship between AP and elderly health. The findings highlight the need to further consider cumulative and life-course AP exposure on elderly health at a smaller geographical scale.

The potential of image-guided orthopaedic navigation to improve surgical outcomes has been well-recognised during the last two decades. According to the tracked pose of target bone, the anatomical information and preoperative plans are updated and displayed to surgeons, so that they can follow the guidance to reach the goal with higher accuracy, efficiency and reproducibility. Despite their success, current orthopaedic navigation systems have two main limitations: for target tracking, artificial markers have to be drilled into the bone and calibrated manually to the bone, which introduces the risk of additional harm to patients and increases operating complexity; for guidance visualisation, surgeons have to shift their attention from the patient to an external 2D monitor, which is disruptive and can be mentally stressful. Motivated by these limitations, this thesis explores the development of an intuitive, compact and reliable navigation system for orthopaedic surgery. To this end, conventional marker-based tracking is replaced by a novel markerless tracking algorithm, and the 2D display is replaced by a 3D holographic Optical see-through (OST) Head-mounted display (HMD) precisely calibrated to a user's perspective. Our markerless tracking, facilitated by a commercial RGBD camera, is achieved through deep learning-based bone segmentation followed by real-time pose registration. For robust segmentation, a new network is designed and efficiently augmented by a synthetic dataset. Our segmentation network outperforms the state-of-the-art regarding occlusion-robustness, device-agnostic behaviour, and target generalisability. For reliable pose registration, a novel Bounded Iterative Closest Point (BICP) workflow is proposed. The improved markerless tracking can achieve a clinically acceptable error of 0.95 deg and 2.17 mm according to a phantom test. OST displays allow ubiquitous enrichment of perceived real world with contextually blended virtual aids through semi-transparent glasses. They have been recognised as a suitable visual tool for surgical assistance, since they do not hinder the surgeon's natural eyesight and require no attention shift or perspective conversion. The OST calibration is crucial to ensure locational-coherent surgical guidance. Current calibration methods are either human error-prone or hardly applicable to commercial devices. To this end, we propose an offline camera-based calibration method that is highly accurate yet easy to implement in commercial products, and an online alignment-based refinement that is user-centric and robust against user error. The proposed methods are proven to be superior to other similar State-of- the-art (SOTA)s regarding calibration convenience and display accuracy. Motivated by the ambition to develop the world's first markerless OST navigation system, we integrated the developed markerless tracking and calibration scheme into a complete navigation workflow designed for femur drilling tasks during knee replacement surgery. We verify the usability of our designed OST system with an experienced orthopaedic surgeon by a cadaver study. Our test validates the potential of the proposed markerless navigation system for surgical assistance, although further improvement is required for clinical acceptance.

Drawing from conceptual discussions of museums over the past five decades, this thesis investigates a newly emergent global trend for institutions to identify themselves as 'cultural hubs', in particular via a case study of Design Society in Shenzhen, China. Literatures in museum studies and visitors studies were reviewed in developing a theoretical framework, and the development of Chinese museums was also included for a contextual understanding of the case study. Surveys, interviews, document analysis, observations and autoethnography were carried out to examine the concept and practices of Design Society and its visitors' experiences. As a new institution, Design Society tries to transcend traditional boundaries associated with museums by identifying itself as a cultural hub. This study demonstrates that cultural programmes at Design Society provide multidimensional experiences for visitors. The object experiences, sensory experiences, transformative experiences, physical experiences and social experiences had by visitors demonstrate the desired characteristics of a cultural hub which is visitor-centred, experience-driven, and where opportunities for visitors to interact, co-create and share are considered important. However, as Design Society struggled to articulate and communicate its identity, 'cultural hub' was perhaps not the most helpful concept for them to utilise, especially in a context where historically narrow definitions of a museum dominate. The development of institutions like Design Society may have a profound impact on museum theory and practice, although it is still too early to judge whether they dilute or serve to reimagine the notion of a museum, at a time when that notion itself is in flux. Nevertheless, this thesis deepens our understanding of visitor experiences, testing a revised model for exploring its parameters within these contexts.

Extended Reality (XR), including augmented (AR), mixed (MR), and virtual reality (VR), is recognized as the key application for next-generation network services. The major service device, head-mounted display (HMD), is also highly expected to continue powering the growth of the consumer electronic market after the personal mobile phone. The internal motivation for such growth is that XR extends the edge of imagination and revolutionize the interaction between human and information by integrating virtual information with the real world. Thus, to fully explore the potential of XR services, enabling mobility through the cellular network is a definite must, which allows the XR service to reach users anytime and anywhere. However, due to the crucial bandwidth and latency requirements, the current XR devices either require a wired connection for high-resolution and low latency video transmission or perform on-device rendering with limited battery life and image quality. The wireless XR schemes, especially cellular networks, struggle to fulfill all these requirements. The lack of mobility support blocks the massive marketization of mobile XR. This thesis features a comprehensive study of the background study and enablers for mobile XR service, especially providing solutions for mobility with the support of machine learning methods. The main contribution of this thesis includes: quality-of-service (QoS) definition for mobile XR services, enhancing the reliability of cellular XR by cooperative transmission, optimizing the cooperate decisions with federated distributed deep reinforcement learning, hierarchical deep reinforcement learning for joint optimization of scheduling and cooperation problems, and the theoretical analysis of the federated multi-agent algorithm. The thesis firstly explains and shows background studies of QoS definitions for different XR use cases under current technical stages from the aspect of wireless communication. In literature, XR QoS metrics are mostly concluded as high bandwidth and low latency, which are oversimplified and too crucial. Adopting these metrics usually results in discussions of mmWave, optical, or Terahertz transmission without the hope of commercialization. Thus, before investigating the enabling technologies of mobile XR, I need to deeply and clearly define the XR QoS. To study the definition of QoS metrics, I built several testbeds with state-of-the-art devices. I built a fourth-generation cellular network testbed with soft-defined radio devices for several popular XR use cases, including unmanned vehicle control and XR conference applications. We measure their requirements and performance. We also build a tiled-DASH broadcast system to investigate the characteristics of XR 360° video traffic. Based on testbeds' results, we show that XR service generally requires consistently low-latency and high bandwidth connection, which is defined as a combination of ultra-reliable low latency communications and Enhanced Mobile Broadband traffic by standards. However, we also notice that the QoS metrics can largely vary for different use cases. Given the example of social media, the users request video messages on transportation with limited interaction contents and highly tolerable to low bandwidth and high latency connection. Thus, it is still possible for cellular networks to concentrate resources on metrics that matter and provide XR services that vary between use cases with the help of certain cellular networks or media technologies. In the sequel, I first recognized that the key point for mobile XR is reliability with testbed results. Unlike conventional cellular services, the jitter or glitch can immediately bring QoS down for XR services. In conventional video streaming, the users can easily load minutes of future resources into their buffer when connection allows in case of sudden connection loss. But it is not possible for XR services with motion correlated resources requests. Thus, the network must provide seamless service for XR users during the entire service duration. Although in the white paper on fifth-generation (5G) cellular networks, the network should constantly offer at least 50Mbps bandwidth. The blind spot still exists in the cell-edge area caused by interference or blockage. To extend the coverage and ensure QoS for XR services, it is possible to group neighboring cells together and jointly serve the users in the cell-edge area. Technologies like Coordinated Multipoint and cell-free MIMO are introduced to standard following this idea. However, the complexity of searching for optimum grouping decisions grows exponentially with the number of participating cells, i.e. dimension explosion. Current solutions, such as greedy or game theorem methods, can only obtain sub-optimal decisions but with relatively low complexity. They either fail in local optimal or require strong assumptions. In this part of the thesis, I applied deep reinforcement learning (DRL) algorithms to solve this problem, which is proved effective for many wireless communication problems. The algorithm is capable to learn, optimize, and adjust the action policy following the environment observations and corresponding reward. The capability of such algorithms is further extended with the support of the neural network and has been shown to outperform many conventional algorithms. But the dimension explosion still limiting DRL, I discussed the distributed deep reinforcement learning, which allows each base station to make its own decisions by optimizing local policy with the local environment and neighborhood decisions, thus significantly reducing the dimensions of each base station's problem. The algorithm then suggests the grouping decision based on the features extracted from the surrounding environment of the single base station. But such features can be inaccurate due to the non-stationary environment caused by unknown neighbors' policies, which limits the performance of distributed deep reinforcement learning. Such a non-stationary environment de-stabilizes the learning procedure and increases the difficulty of convergence. To further enhance the performance, I tried federated learning methods that allow the information (intention and knowledge) exchange among the cells by sharing parameters among the network. I proved that federated learning methods can significantly improve the stability and convergence speed of distributed deep reinforcement learning. My simulation results show-case that the federated distributed deep reinforcement learning can outperform the conventional methods. Besides grouping decisions, there exist many other optimization problems in the transmission pipeline, for example, scheduling. These problems can correlate to each other, while individually optimizing each problem may result in sub-optimal solutions. In the third part of this thesis, I considered a joint optimization of scheduling and cooperative grouping, where the algorithm decides both transmission and re-transmission of video resources and the grouping decision of cooperative transmission. However, the complexity of such an algorithm again increases dramatically for the combined optimization problem. To solve that, I applied hierarchical deep reinforcement learning, which is designed for joint optimization for two-step optimization problems. My results clearly show that it is beneficial to jointly optimize these correlated problems. In the fourth part of my thesis, I performed a comprehensive analysis of the observed gain of federated learning in distributed reinforcement learning in the previous chapter. I re-defined the problem as a networked Markov Decision Process and provide the convergence proof. I also applied the informational model for deep reinforcement learning to measure the information gained during each information exchange. With the results, I quantified the convergence speed and the cost of backhaul traffic during learning, which is proportional to information exchange frequency. My theoretical analysis shows that the exchange of network parameters or experience of considered optimization problem among base stations is useful even with low frequency, which guides efficient algorithm designs and hyper-parameters tuning. Finally, I concluded the thesis with the other mobile XR solutions introduced in the cellular network and some inspirations for possible future works, especially with the reinforcement learning algorithms.

The academic debate on the internationalisation of education has focused on higher education and has overlooked primary and secondary education. The internationalisation of primary and secondary education plays an important role in underpinning internationalisation. English education plays an important role in the global-local networks that have formed to provide educational services which also facilitate individual life chances. This thesis explores the first empirical research undertaken on the internationalisation of primary and secondary education by positioning this topic in debates on international education and business. A qualitative research approach with semi-structured interviews was developed to develop a multiple case study research design. The analysis is based on 98 semi-structured interviews: 40 interviews with English independent schools, 45 interviews with representatives from Translocal Educational Enterprises (TEEs) located in 11 countries, and 13 interviews with stakeholders of English independent schools. A TEE is a business established to provide educational services in a national jurisdiction that combines local investors and other stakeholders with involvement of a non-local school. The research aimed to understand the activities of English independent schools and TEEs. An English independent school becomes involved in a TEE as a non-local school and related brand involved in a school established in another country. A TEE becomes locally embedded but is also embedded in networks linked to an English school. In exploring the establishment processes of the TEEs and the interrelationships between the TEEs and the schools, this thesis makes four contributions. First, it has identified, defined TEEs and outlined their characteristics. The establishment of a TEE reflects a global-local network involving a TEE value network and value proposition that combined global reach with the development of a local value network for the delivery of value propositions for value creation. It has provided an understanding that long-established firms that compete on heritage identity, brand and reputation are involved in a continual process of evolution, alteration and adaptation. Second, it has identified two ways in which primary and secondary education is involved in internationalisation by engaging with Dunning's ownership, location, internalisation (OLI) model (Dunning, 2000, 2009 and 2000) and Hennart's bundling model (Hennart, 2012). It adds value to the bundling models by revealing the ways in which investors from emerging market economies can pay licensing fees to capture firm-specific advantages (FSAs) held by established firms. Given that the TEEs are local educational service business, no entry model concepts are applied to establish a TEE. But investors from emerging market economies have developed four different TEE business models to balance various levels of control with resource commitments as part of a risk reduction return strategy. Third, it has positioned TEEs within the business model and multisided business model literature. In doing so, it has outlined a multisided not-for-profit and for-profit business model framework to enhance the debates on business models. Fourth, it combines the embeddedness and business model literatures by developing an embeddedness-informed business model framework. This framework highlights that the value creation process within an international business is embedded in social relations, including global-local network linkages. Therefore, a TEE reflects some type of continual exchange relationship between places and represents a place-based embeddedness process founded on global-local interactions.

This write-up mainly investigates how the idea of musical kinetic energy and the concept of vectors in physics can be used to extend and analyse principles in music composition, especially in acousmatic music. According to this concept, the methodology for researching the rationale and technology of acousmatic music composition is derived from the concepts of energy and vector, which together provide the foundation of my compositional process. Seven of the submitted works in the portfolio are acousmatic music. They are Xì, Tapping, Cricket, Microwave, If the train goes to Lhasa, Tulpa and Undo. These pieces focus on the intrinsic characteristics of sounds and their extrinsic cultural associations. Because of the properties of different sound samples, each work emphasizes a different type of vector. Three aspects of energy in music composition are expressed in each work: point, line and space. During the development of these works, other music genres have become sources of inspiration for me. These genres have helped develop my methodology and my own unique style of acousmatic music composition. The submitted work in the portfolio – Mandalas – isinstrumental. Compared with acousmatic music, the energy and vectors of instrumental works may be distributed differently in terms of pitch and harmony. Nevertheless, the concepts of energy and vector remain valid for such compositions. In this write-up, I will first introduce the theory of kinetic energy in the field of music. Then, I will discuss the expression of energy in music composition from the three aspects of point, line and space. Directivity and quantity are central to the movement and change of energy in the musical flow, and these concepts are best understood in terms of vectors. Finally, I will discuss seven different types of vectors – gesture, texture-gesture, density, dynamic, frequency, spatial, and meaning, with analysis and examples corresponding to the submitted portfolio.

With the ever-increasing demands for wireless communications from vehicles, vehicle-toeverything (V2X) communications have emerged to enable vehicle user equipment (V-UE) to communicate with other vehicles, pedestrians, and communication infrastructures. In addition, V-UEs are able to offload their computation-intensive applications to a cloud centre or a fog node for processing via a vehicle-to-infrastructure (V2I) link or vehicleto-vehicle (V2V) link. In this case, resource management and offloading optimisation of V2X communications in conjunction with other advanced technologies, such as social networking and cloud/fog computing technologies, are the main paramount challenges for V2X communications underlaying cellular networks. In this thesis, we focus on two different vehicular system models related to the resource allocation and/or the offloading decisions optimisation. The thesis presents three main research contributions. In the first contribution, we investigate how the social attributes of V-UEs in a vehicular network may affect the radio resource allocation mechanism, and the performance of V2X communications. We propose a social-aware clustering-based resource allocation mechanism for V2X communications, which maximises the sum V2I links capacity while guaranteeing the reliability of all V2V links. In the second contribution, we study a vehicular network supported by a mixed cloud and fog computing system, where the queues at the fog node and the cloud centre are modelled following the M/M/1 and M/M/C queueing models, respectively. To minimise the maximum service delay (which includes the transmission delay, the queueing delay and the processing delay) among the V-UEs, we propose to jointly optimise the offloading decisions of all V-UEs and the computation resource allocation at the fog node while considering the V-UEs' mobility and queueing delays at the fog node and the cloud centre. This is achieved by devising a fireworks algorithm-based offloading decision optimisation algorithm in conjunction with a bisection method-based fog node computation resource allocation scheme. In the third contribution, vehicle-carried fog nodes (V-FNs) are deployed in the fog layer and uplink communication resource blocks (RBs) are allocated among all cloud/fog processing V-UEs. We jointly optimise the offloading decisions of all V-UEs, the computation resource allocation at all V-FNs, and the allocation of RB and transmission power for all V-UEs while considering the mobility of V-UEs and V-FNs. The joint optimisation is then solved by devising a fireworks algorithm-based offloading decisions optimisation scheme, in conjunction with a bisection method-based V-FNs computation resource allocation scheme and a clustering-based communication resource allocation scheme.

Subjective wellbeing (SWB), denoting self-reported emotional conditions (e.g. happiness and depressed feelings), is frequently regarded as a critical indicator of quality of life amongst older adults. This thesis focuses on synergising conceptual and methodological innovations in examining the complexity of older Chinese people's SWB regarding how different socioeconomic factors affect their SWB at different (individual and contrxtual) levels.

Measles has been considered a candidate for eradication after smallpox and rinderpest, thanks to the worldwide immunisation campaigns that have been ongoing for more than half a century. However, measles cases have resurged recently, which raises the question of whether the current immunisation strategies, especially the childhood immunisation protocol, are still effective. Furthermore, increasing cases of cross-species transmission of animal morbilliviruses such as canine distemper virus (CDV) make it important to evaluate the level of immunity in the population that would cross-protect against potential zoonotic transmission of morbilliviruses. A pseudotype-based virus neutralisation assay was used to measure neutralising antibody (NAb) titres in serum samples collected from Thai infants at six timepoints before and after two-doses of MMR vaccination to verify the immunogenicity of the childhood immunisation strategy. It was observed that infants born with high levels of maternal immunity did not respond effectively by developing protective titres while infants with lower maternal immunity developed protective titres of antibody following vaccination. It was concluded that increasing the immunogenicity of the first measles immunisation by tailoring the vaccination schedule according to the maternal immunity level acquired by infants could lead to significant improvements in the induction of protective immunity against measles in vaccinated infants. NAb levels were also assessed in a group of sera that were collected from healthcare workers during a measles outbreak in Sheffield in 2016, to indicate the protection status in the "front-line" population. Antibodies against both the vaccine strain and the circulating strains were evaluated to determine the cross-genotypes neutralising activity in the population. The proportion of the samples tested with NAb titres within the "well-protected" range ranged from 82.5% to 93.6% for different genotypes, which fell below the 95% threshold required to eliminate measles. The finding suggested that although cross-neutralisation was observed between the different measles virus (MeV) genotypes tested, a vaccination rate greater than 95% will be required to achieve protective immunity against MeV infection in at least 95% of the population. Furthermore, ELISA were performed in the serum samples using two commercial kits (LIAISON® and EUROIMMUN®) to investigate a possible correlation between the levels of IgG and NAbs. Spearman's rank correlation coefficients were calculated and ranged from 0.40 to 0.55 (P < 0.05) by LIAISON® kit, or 0.71 to 0.79 (P < 0.0001) by EUROIMMUN® kit. The finding demonstrated that it is feasible to use rapid IgG testing as a surrogate measure for neutralising activity to define clinical protection levels within population. Neutralising activities against four morbilliviruses (MeV, RPV, PPRV and CDV) were determined in serum samples collected from either individuals who were previously infected with measles or healthcare workers who had been immunised with measles vaccine. The results indicated that a group of NAbs that can cross-neutralise animal morbilliviruses exist in humans, and the strength of cross-species neutralising activity in humans correlated to the phylogenetic distance between the morbillivirus species. Therefore, maintaining a high MeV vaccination coverage rate across the world is desirable, even after measles eradication. Furthermore, NAbs titres against chimaeric pseudotypes bearing haemagglutinin (H) and fusion (F) proteins from MeV and CDV were tested in human sera, including naturally infected and vaccinated adults and Thai infants. A predominance of MeV-F as a target for NAbs present in both adults and infants was observed when these NAbs were induced by natural infection; however, the predominance of NAb recognising MeV-H was observed in sera collected from vaccinated adults or infants. The results indicated that the induction of NAbs by infection or vaccination could result in different proportion of NAb recognising the H or F glycoproteins and this could impact the cross-neutralisation of animal morbilliviruses. The results of this project suggest that increasing vaccine immunogenicity by targeting vaccination to those susceptible individuals in a population who are able to respond well to vaccination, may be more effective and cost-efficient than vaccinating the whole population indiscriminately. Providing IgG level could correlate with NAb level that is an indicator of clinical protection, it is a promising strategy of using IgG lateral flow test with required cut-off value, which is defined by protective NAb titres, to identify the susceptible individuals rapidly during measles outbreak. Furthermore, the cross-species neutralisation by NAbs recognising the F glycoprotein may be underestimated and warrants further investigation, which could be helpful for preventing the potential zoonotic transmission of animal morbilliviruses, such as CDV to humans.

Space environment is particularly harsh for human activity and human assets. As human activity expands rapidly into space, it becomes ever more urgent to monitor the space environment in situ and in real time. This thesis aims to study a low re-source, miniature detector system to be employed on the satellites to monitor the in-situ space environment particularly for the hazard high energy particles. The vision is that every satellite even at small scale can afford to carry one such detector sys-tem. The core technology used in the detector system is scintillator + SiPM. The scintillator is a mature and traditional radiation detector and the SiPM is the latest generation of photon detector, which rapidly replaces the role of traditional PMT. Development of such detector system requires knowledge over space environment, detector material and geometry, electronics, data processing and more. This thesis tries to have a as complete coverage as possible to demonstrate the flow and the structure of such development. The thesis closely couples the theoretical knowledge to engineering realization. A detector system prototype for pioneering study is fully built and tested from end to end. This serves as the first stepping stone for the path to the full-scale space instrument.

The initial primary motivation driving this research was to test empirical evidence relating to the 'catch up' assumption of EMNEs' strategic asset seeking outwards FDI activities, as per (for example) Luo and Tung (2007)'s springboard perspective and Mathews (2006a)'s LLL model. Specifically, researchers still know very little about whether EMNEs fulfil their firm-level catch-up purposes in terms of post-acquisition innovation performance and how these firms can absorb acquired technologies and other strategic assets, thus leading to acquiring firms' improvement of innovation performance from a combination of resource-based and knowledge-based view with special attention to the impact of home country effects. While it is argued MNEs from emerging markets acquire strategic assets by conducting cross-border M&As with a v view to strengthening their domestic innovation capacity, this presents a paradox, as EMNEs are often considered to lack both the knowledge, resources and absorptive capacity to integrate and exploit such acquisitions by engaging in reverse knowledge transfer. In this thesis, the first part applies resource and knowledge-based views to consider, in particular, how domestic market business group affiliation, prior international experience and state-ownership affect these domestic innovation outcomes. At a theoretical level, factors such as business group affiliation have been considered important and potentially distinguishing characteristics that may make emerging market MNEs somewhat different to developed market MNEs. Empirically, this study applies the propensity score matching (PSM) and difference-in-difference (DID) methods on a sample of 99 international strategic asset seeking motivation M&A deals, plus a control group of 1686 firms. The first section does find evidence of positive impacts of SAS CM&A on patent counts and citations. This first paper also finds firms with international experience and affiliated with business groups perform better, and privately-owned firms outperform state-owned firms in terms of quality of innovation outputs (i.e., patent citations). Application of resource and knowledge-based perspectives helps shed further light on the types of MNEs from emerging markets that may successfully undertake strategic asset seeking acquisitions that may lead to firm-level catch-up. Further, as an additional study and component of this thesis, the second paper explores another question related to Chinese MNE expansion. During the research process a separate but related phenomenon caught the author's attention. When looking at the sample of CM&As, some of the international deals of CMNEs were found to be in unrelated industries. For example, Fosun Group acquired a travel group (Club Med), vi Wanda Group acquired a yacht maker (Sunseeker), while Midea Group acquired an unrelated leader in robotics (Kuka, from Germany). In these cases, the target firms had little relation, in terms of product market characteristics, with the acquiring groups activities. The technology gap between developing and developed market firms is significant and such unrelated product/technologies would make such deals seem even more complicated and difficult to undertake successfully. Indeed, the dominant trend within DMNEs has been moving in the opposite direction: internationally 'horizontal' deals within the same sectors are a growing trait of DMNE internationalization. This has created greater focus on 'core competencies' (UNCTAD, 2013). International divestments of unrelated business, moreover, have led to the dismantling of the remaining iconic Western conglomerates. Siemens and General Electric, for example, as well as lesser-known examples like Maersk, are rapidly shedding non-core businesses (Gapper, 2017). In a second study of this thesis, the author uses event study and buy and hold abnormal return (BHAR) methodology to explore stock market reactions to the international acquisition of unrelated businesses. Interestingly, the results show unrelated international diversification deals perform better than related ones in the long-term but worse in the short-term. Such performance, moreover, is positively modified by diversified business group affiliation and especially in those groups that have R&D capabilities, and mostly magnified when the acquiring firms are affiliated with diversified business groups and at the same time have a strategic-asset-seeking (SAS) motivation. Thus, suggesting that the nature of diversified business group and the SAS motivation play an important role in the success or otherwise of Chinese unrelated international deals. vii In sum, this thesis uses several established methodologies to explore some novel questions that are central to better understanding EMNEs (and Chinese MNEs in particular). In doing so, the thesis contributes to better understanding new theories of emerging market MNE by showing that Chinese MNEs are able to derive benefit from international deals. This suggests that they are able to 'catch up', as stated by Luo and Tung (2007), by acquiring strategic assets in foreign, often developed markets. Being affiliated to a business group supports firms with valuable, rare, inimitable and non-substitutable (VRIN) resources and may provide an excellent environment for organizational learning. Prior international experience, moreover, may provide essential knowledge for acquirers to overcome the liabilities of foreignness they face and achieving performance improvement. While state-owned firms provide sufficient resources, the lack management skills may offset such advantages in improving firm innovation performance. What emerges across both studies is, first, the role of strategic asset seeking motivation. Both studies show an SAS motivation helps increase Chinese acquiring firms' performance, even when the target products are unrelated. Domestic complementary local resources together with the purpose of bringing such strategic assets back for exploitation and development in the domestic market explain why and how Chinese firms apply SAS CM&A to catch-up. In addition, the salient impacts of business group affiliation on post-acquisition performance are highlighted in these studies. The first study shows business group affiliation enhances innovation performance when compared with those non-affiliated firms when they undertake SAS related CM&As. Furthermore, the second study shows that even the deals are unrelated, business group affiliated firms with an SAS orientation perform better. These results shed further light viii on the distinct home country effects within China and how these may influence the ability of Chinese MNEs to exploit and reap value from acquired target firms. In particular, they highlight the apparently important role that business group affiliation may play when it comes to post-acquisition performance. Finally, the negative impacts of state-ownership in both studies point towards the potentially darker side of government involvement in international deals.

With years of researches of biomechanics, biomechanics could be applied to different fields, including medical science, and bionics. Nowadays, there are more problems in human walking movement. Biomechanics could be applied to research how lower limbs work during human walking. In order to figure out how mechanical energy transfers in lower limbs during walking, analysis in different levels would be performed in this thesis. Firstly, gait measurement experiment would be performed to collect static measurement data, dynamic motion data in three different walking speed and external load data with Vicon System and force plate. With smoothing and filtering the raw data with GMAS (General Movement Analysis System), joint kinematics and kinetics could be calculated with motion data and external load data. Afterward, musculoskeletal model would be specified with static measurement data based on the selected exist musculoskeletal model. After building up musculoskeletal model, muscle forces could be computed with motion data and external load data. Thereafter, mechanical energy analysis in joint and segmental level would be performed. The analysis reveals how joints and segments contribute to changing walking speed. The functional behaviours of each joint during different periods of stance phase are determined with function indices. What is more, the mechanical energy transfer between segments would be analysed to determine the energy flow in segments. Finally, mechanical energy analysis in muscle level would be performed. The energy reveals how individual muscle contributes to changing walking speed. The functional behaviours of individual muscle during different periods of stance phase are determined to figure out the functional behaviours' change during stance phase. The mechanical energy via individual muscles are determined with the mechanical work produced or absorbed at insertion and origin points.

Surgery remains the first-line treatment for most cancer types. However, anaesthetic use during surgery may influence cancer cell biology and hence affect cancer recurrence. The current PhD study project objective is to evaluate the effects of two general anaesthetics, sevoflurane (inhalational) and propofol (intravenous), on cancer cells. This study intends to understand the impact of anaesthetics on cancer recurrence following surgery and ultimately provide the research rationale of clinical study/trials for improving cancer surgical outcomes. Propofol inhibited lung, colonic, renal, and ovarian cancer cell malignancy in a dose- dependent manner through regulating PEDF* and HIF-1α expressions. Propofol also altered the metabolites of lung, colonic, renal, and ovarian cancer cells that inhibited the lung and ovarian cancer cell glucose metabolism. It was demonstrated that propofol inhibited lung cancer cell malignancy through downregulating GLUT1 and MPC1, thus disturbing glucose metabolism. These processes induced the high PEDF expression, which downregulated HIF-1α via the Akt pathways, thus regulated pro- and anti-tumour genes. However, propofol neither inhibited brain cancer cell malignancy nor affected the above molecular entities. In contrast, sevoflurane enhanced colonic, renal and ovarian cancer cell malignancy through upregulating VEGFA. Sevoflurane upregulated GLUT1, MPC1 and GLUD1, which enhanced ovarian cancer cellular glucose metabolism. These changes inhibited PEDF expression, and its reduction resulted in upregulating HIF-1α via the Erk pathway, which upregulated two pro-tumour gene-encoded proteins, namely CXCL12 and CXCR4. However, propofol had the opposite effect on these cellular signalling and metabolic pathways in ovarian cancer cells. In summary, this PhD project demonstrated that sevoflurane may have pro-tumour while propofol might have anti-cancer properties. According to laboratory evidence found in the current study, propofol, unlike sevoflurane, may benefit cancer surgery patients. The work presented in this thesis may provide a foundation for further clinical studies to optimise the anaesthesia regimen for better surgical outcomes following cancer surgery.

Controlled Ca²⁺-release from the sarcoplasmic reticulum (SR) during excitation-contraction coupling (EC-coupling) is essential for cardiac and skeletal muscle contraction. Ca²⁺ is released through ryanodine receptors (RyR) in the SR membrane but there are other, different types of ion channels in the SR with unclear physiological roles, and some of these channels have not been identified. In this thesis, I have investigated the structural and biophysical properties of certain ion channels found in the SR, including SR K⁺ channels, trimeric intracellular cation channels (TRIC-A and B) and RyR. The gating and single-channel conductance of SR ion channels was examined after incorporation of vesicles of SR into planar phospholipid bilayers under voltage-clamp conditions. In the search for a ligand that binds to the SR K+ channel, I found that ebselen irreversibly inhibited the channel and that dithiothreitol could partly reverse this effect suggesting the ebselen modified cysteine residues on the channel. It is suggested that TRIC-A or TRIC-B could be the SR K⁺ channel and so I investigated this by characterising the biophysical properties of the ion channels observed after incorporating membrane vesicles obtained from HEK293 cells co-expressing RyR2 plus TRIC-A or RyR2 plus TRIC-B into bilayers. My results indicated that TRIC-A and TRIC-B are non-selective ion channels, permeable to both Cl⁻ and K⁺, and therefore are unlikely to be SR K⁺ channels which are highly selective for monovalent cations. I also found that co-expression of TRIC-A or TRIC-B with RyR2 altered the response of RyR2 to cytosolic Ca²⁺ suggesting that both TRIC-A and TRIC-B interact with RyR2 to regulate RyR2 gating. I established a cardiac-specific Tric-a knockout (Tric-a-cKO) mouse colony as a platform to explore the physiological roles of TRIC channels in the heart. The colony was validated by mRNA and protein levels of other ion channels present in mice hearts. Using homology modelling, I also studied the putative structural changes to RyR2 caused by the mutations V2475F and R2474S that are associated with the lethal arrhythmic disease, catecholaminergic polymorphic ventricular tachycardia type 1 (CPVT1). The predicted models suggested that there are significant conformation changes to the RyR2 helical domain1, thus providing new information regarding the relationship between structural and function in CPVT1.

This thesis is mainly concerned with robust analysis and control synthesis of linear time-invariant systems with polytopic uncertainties. This topic has received considerable attention during the past decades since it offers the possibility to analyze and design controllers to cope with uncertainties. The most common and simplest approach to establish convex optimization procedures for robust analysis and synthesis problems is based on quadratic stability results, which use a single (parameter-independent) Lyapunov function for the entire uncertainty polytope. In recent years, many researchers have used parameter-dependent Lyapunov functions to provide less conservative results than the quadratic stability condition by working with parameterized Linear Matrix Inequalities (LMIs), where auxiliary scalar parameters are introduced. However, treating the scalar parameters as optimization variables leads to large computational complexity since the scalar parameters belong to an unbounded domain in general. To address this problem, we propose three distinct iterative procedures for H2 and H∞state feedback control, which are all based on true LMIs (without any scalar parameter). The first and second procedures are proposed for continuous-time and discrete-time uncertain systems, respectively. In particular, quadratic stability results can be used as a starting point for these two iterative procedures. This property ensures that the solutions obtained by our iterative procedures with one step update are no more conservative than the quadratic stability results. It is important to emphasize that, to date, for continuous-time systems, all existing methods have to introduce extra scalar parameters into their conditions in order to include the quadratic stability conditions as a special case, while our proposed iterative procedure solves a convex/LMI problem at each update. The third approach deals with the design of robust controllers for both continuous-time and discrete-time cases. It is proved that the proposed conditions contain the many existing conditions as special cases. Therefore, the third iterative procedure can compute a solution, in one step, which is at least as good as the optimal solution obtained using existing methods. All three iterative procedures can compute a sequence of non-increasing upper bounds for H2-norm and H∞-norm. In addition, if no feasible initial solution for the iterative procedures is found for some uncertain systems, we also propose two algorithms based on iterative procedures that offer the possibility of obtaining a feasible initial solution for continuous-time and discrete-time systems, respectively. Furthermore, to address the problem of analysis of H∞-norm guaranteed cost computation, a generalized problem is firstly proposed that includes both the continuous-time and discrete-time problems as special cases. A novel description of polytopic uncertainties is then derived and used to develop a relaxation approach based on the S-procedure to lift the uncertainties, which yields an LMI approach to compute H∞-norm guaranteed cost by incorporating slack variables. In this thesis, one of the main contributions is to develop convex iterative procedures for the original non-convex H2 and H∞ synthesis problems based on the novel separation result. Nonlinear and non-convex problems are general in nature and occur in other control problems; for example, the computation of tightened invariant tubes for output feedback Model Predictive Control (MPC). We consider discrete-time linear time-invariant systems with bounded state and input constraints and subject to bounded disturbances. In contrast to existing approaches which either use pre-defined control and observer gains or optimize the volume of the invariant sets for the estimation and control errors separately, we consider the problem of optimizing the volume of these two sets simultaneously to give a less conservative design.

Manufacturers in China face a severe labour shortage as well as labour qualification, which have affected their product quality and productivity severely. One of the causing reasons for that is the new generation of migrant workers who are different from their parent's generation. The new generation of migrant workers seeks not only the right working environment, flexible working time, good salary but also the balance between work and life. The new phenomenon has challenged the way to manage workers in manufacturers in China. This PhD project aims to explore how the application of the internal market orientation, a tool for satisfying the employees' needs while achieving the organisational objectives, can tackle these issues in China. The original internal market orientation construct only covers the needs and wants from the work aspect, this research extends the internal market orientation to include employees' needs and wants in life to achieve a better work-life balance among migrant workers. Following the social exchange theory, work-family border theory, and the theory of planned behaviour, the extended internal market orientation construct is proposed to affect the employees' attitude (employee morale and happiness), and the employees' behaviour (intention to stay and employee productivity). To achieve the proposed research objectives, this study first carried out in-depth interviews with migrant worker and their supervisors to develop new internal market orientation scales that are fit for the Chinese context. Interviews with migrant works are undertaken to identify their needs and wants, while interviews with supervisors are conducted to collect their opinions on the adoption of internal market orientation in the Chinese manufacturing industry. Followed this stage, the qualitative data is embedded in the design of the quantitative survey. The subsequent quantitative study is completed via online questionnaires to examine how the extended internal market orientation affects employees' attitude and behaviour. These results of this study confirm that the extended internal market orientation is a second-order construct constituting internal information collection, internal information dissemination, and internal information responsiveness in work and life, respectively. The extended internal market orientation contributes to improving employee behaviour, in particular, employee productivity when there is a higher level of employee morale and happiness. The supervisor support has a moderating influence on the relationship between the extended internal market orientation and employee happiness rather than employee morale. This study suggests that the organisation should care for employee's psychological wellbeing, which conduces to improving the employees' behaviour and the management should provide and satisfy their needs not only in work but also in life.

This study explores the identity-making of heritage participants and discourse construction of intangible cultural heritage Nüshu (a female created scripts and culture in ancient China, now a national intangible cultural heritage) in contemporary China. This study investigates two heritage discourses in the heritage community: the official heritage discourse and the folk heritage discourse. It demonstrates how participants of these discourses shape, legitimate, and reinforce their heritage identity through different authenticating mechanisms to declare their authenticity in heritage. The theoretical framework of this research links identity, discourse, and heritage, considering the authorised heritage discourse (AHD) and critical heritage studies (CHS) in Chinese context. It also draws on the connection between (intangible) heritage authenticity and cool/hot authentication and how these factors matter heritage identity-making and discourse construction. This study employed multiple methods to obtain the research data, including qualitative in-depth interviews with 25 heritage participants, participant observation in the fieldwork, and a four-year digital ethnography of several digital heritage social media. Critical discourse analysis (CDA) was employed for data analysis. The findings of the research offer insights to the heritage-making and identity-making in the Nüshu ICH community. It shows how Nüshu has become a heritage site of continuing power conflicts among its stakeholders. This study demonstrates how hierarchical authentication and hierarchical identity are made in Chinese intangible cultural heritage. It reveals the relation between "cool" authentication and authenticity declaration and heritage industry construction. The analysis argues the hierarchical, authoritative, and political-economic oriented features of China's official heritage discourse. Then, this study highlights the significance of digitalisation of heritage practices and demonstrates how "hot" authentication is achieved by the folk heritage community through democratic election of folk transmitters on social media. The analysis proves that digital technology and social media have given the non-elites, indigenous members a space to shape and speak for their heritage identity. Heritage practices of the folk participants in virtual community demonstrates a bottom-up, grassroot heritage autonomy with democratic characteristics. This study makes theoretical contributions to Nüshu studies, AHD and CHS theories, and identity-making theories in heritage.

Fire safety is a serious concern when storing a large number of Lithium-ion batteries (LIBs) stacked as an ensemble. Many such fires reported in recent years have caused severe damage to industrial facilities, public property, and even loss of life. It is crucial to understand the mechanisms and causes of these storage fires to provide insights for prevention. While previous studies mostly focused on the chemistry of LIBs and ignition while charging or discharging, this thesis explores the possibility of another fundamental cause of such fires driven by heat transfer, self-heating ignition. Three major challenges are identified for the modelling of self-heating ignition of LIB ensembles: large sizes, multi-dimensional heat transfer, and multiple chemical reactions. In this thesis, a typical LiCoO2 (LCO) battery with four-step reaction kinetics is chosen for analysis and modelling the fundamentals of self-heating ignition. Four numerical models based on COMSOL Multiphysics are developed to deal with these challenges. The numerical results show that the critical ambient temperature triggering self-heating ignition decreases significantly with the size of the battery ensemble, from 155℃ for a single cell to 45℃ for a rack of cells. The spacing and packaging materials used to separate LIBs in storage can promote self-heating ignition further decreasing the critical temperature. The increase in size and the presence of packaging materials result in slower internal heat transfer, which allows the cells to self-ignite at lower ambient temperatures. The heat from self-discharge, which is often neglected in the literature, is predicted to have minor effects on small LIB ensembles but to be dominating for a shelf of LIBs, indicating a substantial change in important chemical mechanism for different sizes. The differences resulting from different numerical models are investigated by a benchmarking analysis using two simulation tools: COMSOL and Gpyro. This thesis provides insights on the fundamental mechanism of self-heating ignition of LIBs during open-circuit storage and scientifically proves that self-heating ignition can be a cause of fires when LIBs are stacked to large sizes.