21684 research outputs found
Sort by
GaitTriViT and GaitVViT: transformer-based methods emphasizing spatial or temporal aspects in Gait Recognition
Get PDFIn image recognition tasks, subjects with long distance and low resolution remains a challenge, whereas Gait Recognition, identifying subjects by walking patterns, is considered one of the most promising biometric technologies due to the stability and efficiency. Previous Gait Recognition methods mostly focused on constructing a sophisticated model structure to better extract spatial and temporal features from frame sequences, aiming to increase the distinctiveness between different feature representations for better model performance during evaluation. Moreover, these methods primarily based on traditional Convolutional Neural Networks (CNNs) due to the dominance of CNNs in Computer Vision.
However, since the alternative form of Transformer, named Vision Transformer, which originally has a wide application in Natural Language Processing (NLP), has introduced into Computer Vision field, the Vision Transformer has gained a strong attention by the outstanding performance in various tasks. Thus, unlike previous methods mainly based on Convolutional Neural Networks (CNNs), this project introduces two Transformer-based method: a completely Vision Transformer-based gait recognition method GaitTriViT and a Video Vision Transformer-based method GaitVViT. The GaitTriViT leveraging Vision Transformer to gain more fine-grained spatial features, while GaitVViT enhances the capacity of temporal extraction. This work evaluates their performances on two of the most popular benchmarks. The results show the still-existing gaps, and several encouraging outperforms compared with current State-of-the-Art (SOTA), demonstrating the difficulties and challenges these Transformer-based methods will encounter continuously. But I still believe in the promising future of Vision Transformers in the field of Gait Recognition
The voter ID debate: Exploring news media and political elite framing of voter ID laws in the US and the UK
Get PDFThis thesis analyses the discourse surrounding voter identification (ID) requirements in the United States (US) and the United Kingdom (UK), focusing on how this controversial voting law is framed by political elites and portrayed in the news.
Over the past decade, conservative elites in the UK and US have passed voter ID requirements (Brennan Center for Justice, 2023b; Elections Act, 2022), on the insistence that the “integrity of elections” needs to be protected. However, voter fraud by impersonation, which is the only form of fraud a voter ID requirement can address, is extremely rare in both countries (The Heritage Foundation, 2023; Electoral Commission, 2022a). Concerns have been raised around the potential impact of these electoral measures on the access to voting of already disadvantaged groups, who are statistically less likely to own the ID required to vote (Combs, 2016; Hicks et al., 2015; Stanford, 2022a; Wang & Nittoli, 2012).
Voter fraud allegations have also gone hand in hand with a decrease in public confidence in elections (Berlinski et al., 2023; Clayton et al., 2021). Despite evidence of voter fraud being extremely rare in both countries, opinion polls show that sections of the British and American public view voter fraud as more common (Electoral Commission, 2022b; Fisher & Sällberg, 2020; Pew Research Centre, 2021). This gap between reality and perceptions is a disconcerting trend for the legitimacy of elections and trust in democratic systems.
This raises the following questions: If there is no available evidence of widespread voter fraud, on what grounds do those who advocate for voter ID laws justify them? What arguments inform the formulation of these electoral reforms? And what is driving differences in public confidence in electoral integrity? While experimental and survey-based studies find political elite’s framing and news consumption to influence public trust and attitudes towards electoral reform, research on these areas, especially in the UK, remains sparse.
To examine political elite rhetoric over voter ID, the key research questions are: (1) How do political elites in the UK and the US frame the issue of voter ID in legislative debates? (2) How does political elite framing of the policy differ based on party affiliations? To examine news coverage of voter ID requirements, as well as to trace temporal changes in news media framing of the issue, the key research questions are: (1) How has the level of attention from news outlets to the issue of voter ID evolved over time? (2) How does the framing of voter ID differ based on the ideological orientation of outlets? (3) Has news framing of voter ID changed over time, i.e. are there any notable shifts in the way these voting requirements are covered?
This thesis is presented in four standalone research papers, each applying a range of computerised content analysis methods to inductively explore language in political debates and news coverage. The first two studies analyse legislative speeches and news media coverage in the US, respectively, and the last two explore Parliamentary debates about voter ID in the UK and coverage by British news media.
The results from the analyses demonstrate attention to this issue has substantially increased and a more complex framing of voter ID requirements than previously theorised. News media is also mirroring elite debates about voter ID and substantially reporting on Donald Trump’s allegations since 2016, particularly left leaning media in the US and also British news outlets. Overall, the findings presented in this thesis help contextualise links identified in the literature between elite framing, individual predispositions and public support for voter ID and provide important insight into key, yet understudied domains shaping public opinion and electoral reform
Authentication enhancement in command and control networks: (a study in Vehicular Ad-Hoc Networks)
Get PDFIntelligent transportation systems contribute to improved traffic safety by facilitating real time communication between vehicles. By using wireless channels for communication, vehicular networks are susceptible to a wide range of attacks, such as impersonation, modification, and replay. In this context, securing data exchange between intercommunicating terminals, e.g., vehicle-to-everything (V2X) communication, constitutes a technological challenge that needs to be addressed. Hence, message authentication is crucial to safeguard vehicular ad-hoc networks (VANETs) from malicious attacks. The current state-of-the-art for authentication in VANETs relies on conventional cryptographic primitives, introducing significant computation and communication overheads. In this challenging scenario, physical (PHY)-layer authentication has gained popularity, which involves leveraging the inherent characteristics of wireless channels and the hardware imperfections to discriminate between wireless devices. However, PHY-layerbased authentication cannot be an alternative to crypto-based methods as the initial legitimacy detection must be conducted using cryptographic methods to extract the communicating terminal secret features. Nevertheless, it can be a promising complementary solution for the reauthentication problem in VANETs, introducing what is known as “cross-layer authentication.” This thesis focuses on designing efficient cross-layer authentication schemes for VANETs, reducing the communication and computation overheads associated with transmitting and verifying a crypto-based signature for each transmission. The following provides an overview of the proposed methodologies employed in various contributions presented in this thesis.
1. The first cross-layer authentication scheme: A four-step process represents this approach: initial crypto-based authentication, shared key extraction, re-authentication via a PHY challenge-response algorithm, and adaptive adjustments based on channel conditions. Simulation results validate its efficacy, especially in low signal-to-noise ratio (SNR) scenarios while proving its resilience against active and passive attacks.
2. The second cross-layer authentication scheme: Leveraging the spatially and temporally correlated wireless channel features, this scheme extracts high entropy shared keys that can be used to create dynamic PHY-layer signatures for authentication. A 3-Dimensional (3D) scattering Doppler emulator is designed to investigate the scheme’s performance at different speeds of a moving vehicle and SNRs. Theoretical and hardware implementation analyses prove the scheme’s capability to support high detection probability for an acceptable false alarm value ≤ 0.1 at SNR ≥ 0 dB and speed ≤ 45 m/s.
3. The third proposal: Reconfigurable intelligent surfaces (RIS) integration for improved authentication: Focusing on enhancing PHY-layer re-authentication, this proposal explores integrating RIS technology to improve SNR directed at designated vehicles. Theoretical analysis and practical implementation of the proposed scheme are conducted using a 1-bit RIS, consisting of 64 × 64 reflective units. Experimental results show a significant improvement in the Pd, increasing from 0.82 to 0.96 at SNR = − 6 dB for multicarrier communications.
4. The fourth proposal: RIS-enhanced vehicular communication security: Tailored for challenging SNR in non-line-of-sight (NLoS) scenarios, this proposal optimises key extraction and defends against denial-of-service (DoS) attacks through selective signal strengthening. Hardware implementation studies prove its effectiveness, showcasing improved key extraction performance and resilience against potential threats.
5. The fifth cross-layer authentication scheme: Integrating PKI-based initial legitimacy detection and blockchain-based reconciliation techniques, this scheme ensures secure data exchange. Rigorous security analyses and performance evaluations using network simulators and computation metrics showcase its effectiveness, ensuring its resistance against common attacks and time efficiency in message verification.
6. The final proposal: Group key distribution: Employing smart contract-based blockchain technology alongside PKI-based authentication, this proposal distributes group session keys securely. Its lightweight symmetric key cryptography-based method maintains privacy in VANETs, validated via Ethereum’s main network (MainNet) and comprehensive computation and communication evaluations.
The analysis shows that the proposed methods yield a noteworthy reduction, approximately ranging from 70% to 99%, in both computation and communication overheads, as compared to the conventional approaches. This reduction pertains to the verification and transmission of 1000 messages in total
Targeted protein degradation of the Y220C-p53 mutant as a potential anti-cancer strategy
Get PDFTargeted protein degradation and Proteolysis Targeting Chimeras (PROTACs) in particular, have become an exciting new field in the drug discovery world, with the potential to target previously seen as ‘undruggable’ proteins. With many PROTACs now reaching the clinic, it is clear that these new modalities have the potential to become therapeutically successful and to overcome problems previously observed with traditional small molecule drugs. One of the most frequently mutated proteins in human cancers, p53 has been a particularly challenging drug target for many years. The work presented in this thesis provides a strategy in which to target and degrade the common Y220C p53 mutant by utilising the PROTAC degradation mechanism.
In particular, PROTAC 2 was found to degrade the Y220C p53 mutant at a low micromolar concentration and hence provides insight into the potential these PROTACs hold for the future as potent degraders of target proteins
Broadband computational rheology for material characterisation
Get PDFRheology is a wide-reaching field whose applications are underpinned by a prior knowledge of the ‘viscoelastic’ properties of (complex) materials generally employed across industries such as oil and gas, food processing, cosmetics, and biophysics; the latter being the main focus of this thesis. Biomedical studies often only have access to small sample volumes, which make conventional bulk rheology techniques unsuitable for their characterization, this has led to the development of a new field called microrheology, where new techniques can characterise the viscoelastic properties of complex fluids by using only a few microlitres of a sample volume. As a branch of rheology, microrheology utilises the same underpinning principles and aims to calculate a material’s properties, including the complex shear modulus, which in turn describes how the material behaves.
The following thesis is aimed at investigating the use of microrheology with optical tweezers in a series of papers exploring different areas within the field of microrheology. Each paper targets certain gaps within the field and as such this thesis is fairly broad reaching touching on algorithm development, machine learning and shear flow analysis.
Chapter 2 presents the work “i-RheoFT: Fourier transforming sampled functions without artefacts”, and introduces an open access MATLAB code, “i-RheoFT”, which can evaluate the Fourier transform of any generic time-dependent function with a finite number of data points. I-RheoFT could be of particular interest and use to those who study sampled or time-averaged functions. The paper investigates three experimental parameters employing i-RheoFT: (i) the density of initial experimental data points that describe the signal, (ii) the interpolation function used to perform virtual oversampling of the signal, which is required for accurate evaluation of the Fourier transform, and (iii) the effect that signal noise has on the Fourier transform. As the chapter shows, a high number of initial data points or a high signal-to-noise ratio corresponds to a good performance for each interpolation function used. Alternatively, a low number of initial points or signal-to-noise ratio corresponds to poor performance across each interpolation function used. As one would expect, there exists a threshold, for both the signal-to-noise and the number of initial points, at which the performance becomes acceptable and has been identified in both cases in the chapter. More recently further development of this work has led to the creation of two open source applications [1, 2] available for download, these aim to compute the complex shear modulus from bulk rheology and atomic force microscopy measurements respectively. Moreover, since its publication this work has been used in three studies [3–5], two of which feature the author of this thesis as a co-author.
Chapter 3 examines the claim that linear microrheology with optical tweezers should not be used for the study of living systems due to the variation between the time required to collect statistically valid data and the mutational time of the studied living system. This work is a first step at enhancing conventional statistical mechanics analysis of particle trajectories, captured using microrheology with optical tweezers, by exploiting machine learning techniques to reduce the current measurement time from tens of minutes down to as little as one second. The chapter describes how computer simulated trajectories, of Newtonian fluids with viscosities spanning three decades, have been used to corroborate the requirement for sufficiently long measurements to offer a good estimation of the fluid viscosity using conventional analytical techniques. In addition, the work provides a method for estimating the measurement time of a microrheology with optical tweezers experiment, based on the relative viscosity of the fluid being analysed to produce an uncertainty as low as 1%. Furthermore, this chapter presents a machine learning algorithm that can predict the viscosity of both simulated and real trajectories, carrying an error as low as ∼ 0.3%, using only one second of data. It is believed that with this machine learning enhancement, microrheology with optical tweezers will become a powerful tool for studies involving living systems.
Chapter 4 presents an investigation into flow induced self-assembly (FISA) of particles suspended in a viscoelastic shear thinning fluid subjected to simple shear flow. This phenomena is currently not fully understood and little has been done in literature so far to investigate the possible effects of the shear-induced elastic instability. In this work, a bespoke cone and plate shear cell is used to provide new insights on the FISA dynamics. In particular, we have fine tuned the applied shear rates to investigate the chaining phenomenon of micron-sized spherical particles suspended into a viscoelastic fluid characterised by a distinct onset of elastic instability. This has allowed us to reveal three phenomena never reported in literature before, i.e.: (I) the onset of the elastic instability is strongly correlated with an enhancement of FISA; (II) particle chains break apart when a constant shear is applied for ‘sufficiently’ long-time (i.e. much longer than the fluids’ longest relaxation time). This latter point correlates well with the outcomes of parallel superposition shear measurements, which (III) reveal a fading of the elastic component of the suspending fluid during continuous shear flows
Assessing the impact of the introduction of a research-led computer science framework for primary education in the kingdom of Saudi Arabia
No full textAbstract not currently available
Fabrication and optimisation of high-performance nanoplasmonic sensors for biosensing applications
Get PDFPlasmonic sensors are established as an effective, label-free technique for many applications. This thesis focuses on localised surface plasmon resonance (LSPR) sensors and how to optimise them for sensing applications. Although there are examples of lithographically-defined nanostructures being deployed in biosensing applications, few of these studies make attempts to use nanostructures optimised for sensing performance. For this reason, in this thesis we examine different ways of optimising plasmonic nanostructures with the aim of improving their performance in biosensing applications.
We first focus on the properties and sensing abilities of different nanostructure geometries including experimentally assessing both the bulk and localised refractive index sensitivity of different nanostructure arrays. This includes a discussion of the merits of different nanostructure geometries for different applications and the important role which resonance wavelength plays in a sensors refractive index sensing capability.
Next, we focus on the optimisation of nanostructure arrays using annealing techniques. There are few studies on the effect of annealing on the sensing performance of nanostructure arrays and this chapter serves as a discussion as to the benefits and drawbacks of thermal annealing for plasmonic nanostructure sensing arrays. As well as discussing the positive effects of annealing, remedies to the main drawback of annealing, structure deformation, are investigated. Structure deformation results in a reduced electromagnetic field concentration, resulting in diminished sensing performance. To mitigate these issues, we explore the use of a protective layer during annealing to prevent deformation as well as the use of a femtosecond laser as an alternative method for annealing.
Finally, we present work towards a cross-reactive method of sensing proteins. This begins by presenting different methods of making the plasmonic sensors reusable after proteins have adhered to the surface including optimising a cleaning protocol and investigating different surface chemistries that may allow for temporary protein interactions. This culminates in initial testing of a cross-reactive system which shows some success in discriminating between different proteins in solution without the need for specific receptors
King John’s military response to the loss of Normandy: The Campaigns of 1206 and 1214
No full textAbstract not currently available
PROTAC-mediated degradation of phospholamban as a novel therapeutic strategy for heart failure
No full textAbstract not currently available
Synthesis and characterisation of Ca-Mg hydrides for solid-state hydrogen storage
No full textAbstract not currently available
Data sources:
Access Repository Dashboard
Do you manage Open Research Online? Become a CORE Member to access insider analytics, issue reports and manage access to outputs from your repository in the CORE Repository Dashboard! 👇