17628 research outputs found
Sort by
FastAdaBelief: Improving Convergence Rate for Belief-Based Adaptive Optimizers by Exploiting Strong Convexity
Get PDFAdaBelief, one of the current best optimizers, demonstrates superior generalization ability over the popular Adam algorithm by viewing the exponential moving average of observed gradients. AdaBelief is theoretically appealing in which it has a data-dependent O(√T) regret bound when objective functions are convex, where T is a time horizon. It remains, however, an open problem whether the convergence rate can be further improved without sacrificing its generalization ability. To this end, we make the first attempt in this work and design a novel optimization algorithm called FastAdaBelief that aims to exploit its strong convexity in order to achieve an even faster convergence rate. In particular, by adjusting the step size that better considers strong convexity and prevents fluctuation, our proposed FastAdaBelief demonstrates excellent generalization ability and superior convergence. As an important theoretical contribution, we prove that FastAdaBelief attains a data-dependent O(log T) regret bound, which is substantially lower than AdaBelief in strongly convex cases. On the empirical side, we validate our theoretical analysis with extensive experiments in scenarios of strong convexity and nonconvexity using three popular baseline models. Experimental results are very encouraging: FastAdaBelief converges the quickest in comparison to all mainstream algorithms while maintaining an excellent generalization ability, in cases of both strong convexity or nonconvexity. FastAdaBelief is, thus, posited as a new benchmark model for the research community
Getting Personal: the issues of trust and distrust in small and medium-sized enterprises in Nigeria
Get PDFThere is a pronounced paucity of empirically rigorous study that focuses on the impact of trust on small and medium-sized enterprises operating in a developing market context. This article therefore offers a fresh perspective on the simultaneous relationship between trust and distrust by exploring the complex process though which they are developed. Constructed in the assumptions of multidimensionality and the inherent tensions of relationships, the design of this study is interpretive, following an emergent iterative process, where three distinct types of trust, cognitive based trust, affect based trust and calculus-based trust were considered as critical components for successful SME relationships. Conversely, the unpredictable negative behaviour of a trade partner was critical to the development of calculus-based distrust and identification based distrust. The results facilitate a better understanding of the distinct types of trust and distrust that underpin SME relationships in Nigeria and other developing economies, particularly in Africa. This article contributes to the ongoing debate over the two contrary yet complementary opposites, and their ability to provide explanations to economic activity
A self-attention integrated spatiotemporal LSTM approach to edge-radar echo extrapolation in the Internet of Radars
Get PDFIn recent years, the number of weather-related disasters significantly increases across the world. As a typical example, short-range extreme precipitation can cause severe flooding and other secondary disasters, which therefore requires accurate prediction of extent and intensity of precipitation in a relatively short period of time. Based on the echo extrapolation of networked weather radars (i.e., the Internet of Radars), different solutions have been presented ranging from traditional optical-flow methods to recent deep neural networks. However, these existing networks focus on local features of echo variations to model the dynamics of holistic radar echo motion, so it often suffers from inaccurate extrapolation of the radar echo motion trend, trajectory, and intensity. To address the problem, this paper introduces the self-attention mechanism and an extra memory that saves global spatiotemporal feature into the original Spatiotemporal LSTM (ST-LSTM) to form a self-attention Integrated ST-LSTM recurrent unit (SAST-LSTM), capturing both spatial and temporal global features of radar echo motion. And several these units are stacked to build the radar echo extrapolation network SAST-Net. Comparative experiments show that the proposed model has better performance on different real world radar echo datasets over other recent methods
Financial independence of women – the impact of social factors on women empowerment in small island developing states (SIDS)
No full textPurposeThis study aims to understand and analyse the financial independence of women in small island developing states, with a focus on Mauritius. Factors such as employer choice, domestic violence, sociological factors, lack of opportunities and empowerment and the legal framework have been identified as potential influencers of the financial independence of women.Design/methodology/approachA survey was conducted where residents of Mauritius were targeted to have a more generic overview of the subject matter. A response rate of 347 was received. The partial least square structural equation modeling was used to analyse the proposed framework.FindingsA total of 12 hypotheses were proposed and only 2 hypotheses were confirmed. The sociological factors, lack of opportunities, domestic violence and employer choice appeared not to have a significant influence on the financial independence of women. The legal system had a significant influence on the financial independence of women.Originality/valueIt must be acknowledged that the literature is rich with studies on financial independence. Nevertheless, not much has been prescribed in the literature from the perspective of small developing economies and having women at the centre of the debate. The theory of gender and power and the social learning theory were used as the theoretical foundation
Evaluating the suitability of close-kin mark-recapture as a demographic modelling tool for a critically endangered elasmobranch population
Get PDFEstimating the demographic parameters of contemporary populations is essential to the success of elasmobranch conservation programmes, and to understanding their recent evolutionary history. For benthic elasmobranchs such as skates, traditional fisheries-independent approaches are often unsuitable as the data may be subject to various sources of bias, whilst low recapture rates can render mark-recapture programmes ineffectual. Close-kin mark-recapture (CKMR), a novel demographic modelling approach based on the genetic identification of close relatives within a sample, represents a promising alternative approach as it does not require physical recaptures. We evaluated the suitability of CKMR as a demographic modelling tool for the critically endangered blue skate (Dipturus batis) in the Celtic Sea using samples collected during fisheries-dependent trammel-net surveys that ran from 2011 to 2017. We identified three full-sibling and 16 half-sibling pairs among 662 skates, which were genotyped across 6291 genome-wide single nucleotide polymorphisms, 15 of which were cross-cohort half-sibling pairs that were included in a CKMR model. Despite limitations owing to a lack of validated life-history trait parameters for the species, we produced the first estimates of adult breeding abundance, population growth rate, and annual adult survival rate for D. batis in the Celtic Sea. The results were compared to estimates of genetic diversity, effective population size (Ne), and to catch per unit effort estimates from the trammel-net survey. Although each method was characterized by wide uncertainty bounds, together they suggested a stable population size across the time-series. Recommendations for the implementation of CKMR as a conservation tool for data-limited elasmobranchs are discussed. In addition, the spatio-temporal distribution of the 19 sibling pairs revealed a pattern of site fidelity in D. batis, and supported field observations suggesting an area of critical habitat that could qualify for protection might occur near the Isles of Scilly
Is it possible to differentiate ICD-11 complex PTSD from symptoms of borderline personality disorder?
Get PDFThe relationship between the Dark Triad and attitudes towards feminism
Get PDFThe Dark Triad traits are considered a male-centric framework of personality with women generally scoring lower on narcissism, Machiavellianism, and psychopathy. Research has examined the drivers behind this relationship attributing effects mostly to biological or evolutionary reasons with less work understanding environmental factors. To date, no research has examined the relationship between the Dark Triad and attitudes towards feminism. Three hundred and forty-three participants completed self-report measures of the Dark Triad and feminist attitudes. Results reported no differences between men and women on feminist attitudes, but men scored higher on the Dark Triad. Multiple linear regression indicated a negative association between the Dark Triad and feminist attitudes with all three traits significantly negatively contributing to the model. In all cases, this effect was stronger in men. These findings suggest that whilst men and women hold similar feminist attitudes, Dark Triad traits may facilitate a disregard for feminism
A professional money laundering scandal; a narrative-based exploration of undocumented foreign workers in a large construction project
No full textPurposeMoney laundering weakens the role of the construction industry in stimulating economic growth. The purpose of this paper is to explore the connection between money laundering on the construction sites and undocumented foreign workers, based on a narrative drawn from a qualitative research.Design/methodology/approachThroughout the study, qualitative methods, i.e. interviews, site visits and document analysis, were used. However, the data for this paper was primarily derived from an interview. Thematic analysis was used to analyse the data.FindingsThe findings show that construction personnel who have access to the business’s financial affairs are the most likely to engage in illicit transactions. The size of the project as well as the multiple layers of organisations involved made it easy for launderers to operate. The appealing commission provided incentives to opportunistic personnel. In this regard, the wages for undocumented workers, which were primarily paid in cash, provided a considerable opportunity for the subcontracting organisations to engage in money laundering.Research limitations/implicationsWhile the single narrative method with an omniscient narrator allows for the conceptualisation of a human experience with money laundering, the depth of information and interpretations is limited. Emerging qualitative research methods may be incorporated in the future to provide a more extensive information due to the fact that money laundering data is complex and sensitive that few people want to discuss.Originality/valueThe multidisciplinary approach of this research provides a pedagogical way that focuses primarily on the disciplines of construction management and business ethics to demonstrate real-world money laundering practice. Understanding such phenomenon on sites opens up key avenues for future research into developing an anti-money laundering regime for the construction industry
Scenario-based incident response training: lessons learnt from conducting an experiential learning virtual incident response tabletop exercise
No full textPurposeThis paper aims to discuss the experiences designing and conducting an experiential learning virtual incident response tabletop exercise (VIRTTX) to review a business's security posture as it adapts to remote working because of the Coronavirus 2019 (COVID-19). The pandemic forced businesses to move operations from offices to remote working. Given that this happened quickly for many, some firms had little time to factor in appropriate cyber-hygiene and incident prevention measures, thereby exposing themselves to vulnerabilities such as phishing and other scams.Design/methodology/approachThe exercise was designed and facilitated through Microsoft Teams. The approach used included a literature review and an experiential learning method that used scenario-based, active pedagogical strategies such as case studies, simulations, role-playing and discussion-focused techniques to develop and evaluate processes and procedures used in preventing, detecting, mitigating, responding and recovering from cyber incidents.FindingsThe exercise highlighted the value of using scenario-based exercises in cyber security training. It elaborated that scenario-based incident response (IR) exercises are beneficial because well-crafted and well-executed exercises raise cyber security awareness among managers and IT professionals. Such activities with integrated operational and decision-making components enable businesses to evaluate IR and disaster recovery (DR) procedures, including communication flows, to improve decision-making at strategic levels and enhance the technical skills of cyber security personnel.Practical implicationsIt maintained that the primary implication for practice is that they enhance security awareness through practical experiential, hands-on exercises such as this VIRTTX. These exercises bring together staff from across a business to evaluate existing IR/DR processes to determine if they are fit for purpose, establish existing gaps and identify strategies to prevent future threats, including during challenging circumstances such as the COVID-19 outbreak. Furthermore, the use of TTXs or TTEs for scenario-based incident response exercises was extremely useful for cyber security practice because well-crafted and well-executed exercises have been found to serve as valuable and effective tools for raising cyber security awareness among senior leadership, managers and IT professionals (Ulmanová, 2020).Originality/valueThis paper underlines the importance of practical, scenario-based cyber-IR training and reports on the experience of conducting a virtual IR/DR tabletop exercise within a large organisation
Data sources:
Access Repository Dashboard
Do you manage Repository@Napier? Access insider analytics, issue reports and manage access to outputs from your repository in the CORE Repository Dashboard!