30793 research outputs found
Sort by
A Lightweight Object Counting Network Based On Density Map Knowledge Distillation
Get PDFObject counting aims to count the accurate number of object instances in images, and its operation efficiency is essential. However, most current CNN-based methods rely on complex network architectures, which results in them consuming a significant amount of memory, time, and other resources at runtime. This seriously limits their deployment in practical application scenarios, such as public safety and agriculture planting. Therefore, we propose a lightweight object counting method named EdgeCount to effectively balance inference speed and object counting accuracy. Specifically, we construct a network composed of a student model (EdgeCount) and a teacher model (EdgeCount-T) with the same encoder-decoder structure based on density map knowledge distillation (DMKD), allowing the EdgeCount to learn object density distribution from the EdgeCount-T. After that, we introduce spatial and channel reconstruction convolution (SCConv), composed of a spatial reconstruction unit (SRU) and a channel reconstruction unit (CRU), to decrease spatial and channel redundancy with lower computational costs. Moreover, a low parameter weighted multi-scale feature fusion module (LWMFFM) is designed to further improve the countering ability through segmenting minor structural discrepacies among multi-scale features. Extensive experiments conducted on challenging remote sensing and dense crowd object counting datasets demonstrate the effectiveness and superiority of our method. In particular, under the four NVIDIA Jetson devices, EdgeCount can accurately counter objects with only 0.12M parameters and 19.87M floating-point operations per second (FLOPs) in the size of 128, which achieves the lowest latency and fastest FPS compared with other state-of-the-art object counters.10.13039/501100005230-Natural Science Foundation of Chongqing Municipality (Grant Number: Grant No. cstc2020jcyj-cxttX0002);
10.13039/501100001809-National Natural Science Foundation of China (Grant Number: Grant No. U21A20447)
Shaping ESG commitment through organizational psychological capital: The role of CEO power
Get PDFData Availability Statement: Data available on request from the authors.This study investigates the influence of organizational psychological capital (OPC) on corporate environmental, social, and governance (ESG) practices, highlighting a relatively overlooked aspect in existing studies, and examines the moderating effect of chief executive officer (CEO) power on this relationship. Using a dataset of 1659 firm-year observations from FTSE 350 firms across the years 2012–2021 and applying natural language processing (NLP) techniques, our findings reveal that higher levels of OPC are linked to a stronger commitment to ESG initiatives. However, this positive association is tempered by CEO power, which negatively moderates the relationship. Furthermore, our analysis shows that OPC not only enhances ESG performance but also positively influences financial performance and the core ESG pillars. These results, validated through rigorous robustness checks, offer significant insights for stakeholders and policymakers in the realm of corporate governance
Disability on the Move: Disabled Mobilities in Contemporary India
Get PDFIn this chapter, we explore multiple constellations of disability to challenge the binary opposition drawn between stasis and mobility, and to document the complex ways in which the two are entangled. We draw on multi-sited fieldwork with people affected by leprosy in coastal South India; Indian Sign Language (ISL)-speaking deaf young adults in urban locations across India;
and others identified by themselves or others as disabled in Hyderabad, to document the journeys—for treatment, education, work, and activism—taken by disabled people across their life courses. While these movements should not be conflated with progress, we argue, they do create particular creative opportunities for people to transcend identities and life projects and to forge new ones that are not always open to their nondisabled peers
When atypical leaders fail to deliver allyship for diversity: The case of an unregulated neoliberal national context
Get PDFOrganisations increasingly embrace allyship as a strategy to enhance support for diversity. The rise of atypical leaders offers hope to individuals from marginalised backgrounds, fostering the belief that these leaders would align themselves as allies and actively promote diversity within organisations. However, this assumption remains empirically untested. This paper investigates the tendency of atypical leaders to engage in allyship behaviours in contexts where regulatory and normative support for diversity is absent. Within unregulated neoliberal environments, the significance of atypical leaders is amplified, as diversity initiatives frequently receive limited backing from typical leaders, and the lack of a regulatory framework subjects these initiatives to considerable strain and risk. Through a qualitative study involving 33 atypical leaders from Turkey, we explore whether atypical leaders exhibit allyship towards diversity. Our findings delineate the conditions that enable and limit the effectiveness of atypical leaders’ allyship in a country with a toxic triangle of diversity. This study illuminates the critical influence of the regulatory environment on the allyship behaviours of atypical leaders, underlining the complex interplay between leadership, regulatory contexts, and allyship practices.The author(s) received no financial support for the research, authorship, and/or publication of this article
Using natural driving experiments and Markov chains to develop realistic driving cycles
Get PDFData availability: The data that has been used is confidential.The main purpose of driving cycles is to estimate accurately on-road fuel use and the associated emissions of greenhouse gases and other air pollutants by vehicles. Conventionally, driving cycles are developed using micro-trips, Markov chains, or hybrid approaches, with accuracy determined by comparing metrics of the candidate cycles with the observed data. Through a natural driving experiment, we suggest traffic and road topology have a dominant role in influencing individual driving styles, more so than driver age or gender, or vehicle characteristics. Using experimental data and a Markov chain approach, we make three contributions to driving cycle development. First, we identify a reduced set of 26 metrics which materially influence fuel economy. Second, we assess the trade-offs in accuracy between reproducing vehicle dynamics and fuel economy. Finally, we identify the impact of natural driving variability on the accuracy of candidate cycles
Determinants of health-seeking behaviour among internal migrants in Ghana: A study of the North South migration
Get PDFThis thesis was submitted for the award of Doctor of Philosophy and was awarded by Brunel University LondonBackground: Healthcare access and utilisation among internal migrants in Ghana present complex challenges influenced by various social, economic, and structural factors. Addressing these challenges requires a comprehensive understanding of the barriers and facilitators shaping health-seeking behaviour within this population. This study aims to provide such insights by thoroughly examining the determinants of health-seeking behaviour among internal migrants in Ghana.
Methods: This study utilised multiple approaches to answer the research questions. These approaches included a literature review, two quantitative analyses using regressions, two qualitative analyses involving a Delphi, and in-depth interviews. A systematic review was initially conducted to identify the gaps in the literature regarding the determinants of health-seeking behaviour in Africa. Regression analyses were then conducted to identify the barriers and facilitators of health-seeking behaviour among internal migrants in Ghana using the Ghana Living Standard Survey Round 7 (GLSS7) dataset from the Ghana Statistical Service (GSS). Using the same dataset, regressions were conducted to measure the impact of healthcare costs on the factors influencing healthcare service utilisation among this vulnerable population. Delphi study to solicit information on the perspective of healthcare deliverers regarding internal migrants’ health-seeking behaviour in Ghana was conducted. Finally, in-depth interviews were conducted to identify the socio-cognitive perceptions regarding internal migrants’ health-seeking behaviour in Ghana.
Results: The review revealed gaps in the literature regarding migrant healthcare. The analyses revealed an 8% lower healthcare service utilisation rate among internal migrants, highlighting existing access barriers. Self-care practices were prevalent, with more than 90% of respondents relying on them. Demographic factors significantly influenced healthcare utilisation: individuals aged 18-35 and females exhibited higher utilisation rates. Enabling factors like health insurance and income showed mixed associations, while financial capability strongly influenced healthcare seeking. There was no significant relationship between illness type and service utilisation. Determinants of out-of-pocket healthcare expenditure included age, education, location, marital status, and place of seeking healthcare. Barriers identified through the Delphi study included appointment wait times, language difficulties, and financial constraints, while insurance possession and higher income levels facilitated healthcare access. Policy recommendations focused on active insurance policies and education. Awareness levels varied, with 40% considering healthcare crucial, and coping strategies included reliance on support networks and self-care practices.
Conclusion: This study meticulously addresses the barriers and facilitators of health-seeking behaviour among internal migrants in Ghana through four well-defined objectives. Integrating quantitative and qualitative methodologies enhances the exploration, offering nuanced insights into healthcare access and utilisation. The findings contribute significantly to healthcare discourse, providing a reliable foundation for targeted interventions, policy improvements, and future research endeavours, ultimately fostering healthcare equity for internal migrants in Ghana
Exploring 8-bit Arithmetic for Training Spiking Neural Networks
Get PDFSpiking Neural Networks (SNNs) offer advantages over traditional Artificial Neural Networks (ANNs) in terms of biological plausibility, noise tolerance, and temporal processing capabilities. Additionally, SNNs can achieve significant energy efficiency when deployed on specialized neuromorphic platforms. However, the common practice of training SNNs using Back Propagation Through Time (BPTT) on GPUs before deployment is resource-intensive and hinders scalability. Although reduced precision inference with SNNs has been explored, the use of reduced precision during training remains largely unexamined. This study investigates the potential of posit arithmetic, a novel numerical format, for training SNNs on future posit-enabled accelerators. We evaluate the performance of 8-bit posit and floating-point arithmetic compared to 32-bit floating-point on two datasets. Our results show that 8-bit posits can match the performance of 32-bit floating-point arithmetic when all training components are quantised. These findings suggest that posit arithmetic could be a promising foundation for developing efficient hardware accelerators dedicated to SNN training. Such advancements are essential for reducing resource usage and enhancing energy efficiency, enabling the exploration of larger and more complex SNN architectures, and promoting their wider adoption.10.13039/501100000266-EPSRC (Grant Number: EP/V052241/1,EP/S030964/1
CHUNAV: Analyzing Hindi Hate Speech and Targeted Groups in Indian Election Discourse
Get PDFIn the ever-evolving landscape of online discourse and political dialogue, the rise of hate speech poses a signiicant challenge to maintaining a respectful and inclusive digital environment. he context becomes particularly complex when considering the Hindi language—a low-resource language with limited available data. To address this pressing concern, we introduce the CHUNAV dataset—a collection of 11,457 Hindi tweets gathered during assembly elections in various states. CHUNAV is purpose-built for hate speech categorization and the identiication of target groups. he dataset is a valuable resource for exploring hate speech within the distinctive socio-political context of Indian elections. he tweets within CHUNAV have been meticulously categorized into “Hate” and “Non-Hate” labels, and further subdivided to pinpoint the speciic targets of hate speech, including “Individual”, “Organization”, and “Community” labels (as shown in Figure 1). Furthermore, this paper presents multiple benchmark models for hate speech detection, along with an innovative ensemble and oversampling-based method. he paper also delves into the results of topic modeling, all aimed at efectively addressing hate speech and target identiication in the Hindi language. his contribution seeks to advance the ield of hate speech analysis and foster a safer and more inclusive online space within the distinctive realm of Indian Assembly Elections.MKis supported by UKRI NERC grant NE/X000192/12
Analyzing the quality of service of quantum oriented open radio access networks in 5G and 6G
Get PDFThis thesis was submitted for the award of Doctor of Philosophy and was awarded by Brunel University LondonThe Open Radio Access Network (ORAN) technology has been developed to provide
efficient spectrum sharing and cost-effective solutions, but it also increases network
infrastructure and power usage. To evaluate the performance of the power
consumption (PC), a model was proposed to quantify the compromises associated
with virtualizing a server within an ORAN infrastructure, considering factors like the
quantity of virtual machines, allocation of system resource blocks, and bandwidth.
However, virtualization of the network has resulted about 50% reduction in the total
PC in comparison with traditional cloud networks. However, the ORAN paradigm
has produced more PC compared to the virtualization case, about 30% in the total
PC and 10% in the cooling PC. Unless the advantages of ORAN are fully realized,
the addition of extra units within the ORAN, specifically the DU servers, may result
in more PC that might advocate against the ORAN.
Subsequently, a work was proposed to examine the criteria and evaluations for
prospective quantum solutions in conventional ORAN networks, focusing on entanglement
phenomena to enhance the efficiency of the X2 application (X2-AP) protocol.
This approach reduces the overhead of X2-AP signaling, reducing time and power consumption
associated with standard cloud-based systems. As a result, increasing the
number of photons has decreased the delay to about 40% compared to the traditional
ORAN network. In addition, the energy efficiency in the quantum case has been
increased while decreasing the power consumption by about 10%.
This study also investigates the use of quantum entanglement-based approaches
and their influence on conventional ORAN architecture’s signaling among the central
units (CUs) and distributed units (DUs) by replacing the traditional method with
entangled photons. The results showed that the proposed method has promised about
45%, 40% and 10% reductions in the EE, PC and delay, respectively, when compared
to the traditional ORAN. Finally, a novel methodology for addressing problems in ORAN and implementing
load balancing algorithms is presented, focusing on selecting ORAN servers with
lower energy efficiency for quantum load balancing. This comparative analysis offers
valuable insights for developing power-efficient ORAN implementations. This
nonlinear problem was solved using Lagrange multiplier method to solve the problem
mathematically, and using the Matlab (fmincon) software to solve the problem
numerically. In which, two algorithms are used, sequential quadratic programming
(SQP) and active-set. It was shown that the SQP model exhibits superior energy
efficiency compared to the active-set model, with a difference of approximately 45%.Ministry of Higher Education and Scientific Research (MOHESR), Cultural Attache, and University of Diyala in Ira
Taxonomic insights into ethereum smart contracts by linking application categories to security vulnerabilities
Get PDFData availability: The datasets generated and/or analysed during the current study are available in the Top-Trending-Contracts repository, https://github.com/giacomofi/Top-Trending-Contracts. To support ongoing research and community engagement, we have established a GitHub repository, currently featuring a collection of the top-trending contracts categorized as ELTC, Bank, CNFT, and Token programs, organized chronologically. While the repository presently focuses on these trending categories, our plan is to progressively include a broader spectrum of contracts, particularly those representing other categories and various vulnerabilities. This expansion will not only improve the repository’s but also provide insights into the vulnerability landscape of Ethereum smart contracts.The expansion of smart contracts on the Ethereum blockchain has created a diverse ecosystem of decentralized applications. This growth, however, poses challenges in classifying and securing these contracts. Existing research often separately addresses either classification or vulnerability detection, without a comprehensive analysis of how contract types are related to security risks. Our study addresses this gap by developing a taxonomy of smart contracts and examining the potential vulnerabilities associated with each category. We use the Latent Dirichlet Allocation (LDA) model to analyze a dataset of over 100,040 Ethereum smart contracts, which is notably larger than those used in previous studies. Our analysis categorizes these contracts into eleven groups, with five primary categories: Notary, Token, Game, Financial, and Blockchain interaction. This categorization sheds light on the various functions and applications of smart contracts in today’s blockchain environment. In response to the growing need for better security in smart contract development, we also investigate the link between these categories and common vulnerabilities. Our results identify specific vulnerabilities associated with different contract types, providing valuable insights for developers and auditors. This relationship between contract categories and vulnerabilities is a new contribution to the field, as it has not been thoroughly explored in previous research. Our findings offer a detailed taxonomy of smart contracts and practical recommendations for enhancing security. By understanding how contract categories correlate with vulnerabilities, developers can implement more effective security measures, and auditors can better prioritize their reviews. This study advances both academic knowledge of smart contracts and practical strategies for securing decentralized applications on the Ethereum platform
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! 👇