6974 research outputs found
Sort by
Self-Organising Explainable Multi-View Representation Learning for Remote Sensing Scene Classification
No full textRemote sensing scene classification is widely considered to be a challenging task due to high intraclass variability and interclass similarity in remotely sensed imagery. While existing deep neural networks achieve promising performance, they often lack transparency and generalisation capability. To enhance interpretability without sacrificing accuracy, a novel self-organising transparent multi-view representation learning framework based on evolving fuzzy neural encoders for remote sensing scene classification is introduced in this paper. The framework leverages multiple pre-trained convolutional neural network backbones with different architectures to extract image embeddings from multiple views. The multi-view image embeddings are projected into a lower-dimensional feature space using multilayer evolving fuzzy neural networks trained in a supervised or self-supervised fashion as encoders and subsequently fused for scene classification. Extensive experiments on six benchmark datasets (Optimal-31, WHU-RS, UCMerced, AID, RSI-CB256, and PatternNet) demonstrate the framework’s superior performance, achieving average accuracies of 99.81 %, 98.83 %, 97.86 %, 98.37 %, 99.84 %, and 98.83 %, respectively, without fine-tuning to the specific context. Ablation studies confirm the complementary contributions of the multi-view, supervised, and self-supervised components in the proposed framework. The proposed framework provides an effective solution for remote sensing scene classification, achieving high accuracy with enhanced transparency and interpretability
Public perspectives on the 20 mph speed reduction in Wales:A Latent Dirichlet Allocation Approach
No full textINTRODUCTION: This study investigates public attitudes toward the Welsh Government's implementation of a 20 mph default speed limit on all non-unrestricted roads, for example those in urban environments, residential neighborhoods, and areas with high pedestrian activity, that previously were 30 mph, aimed at reducing traffic crashes, emissions, and improving quality of life.METHOD: Using a Latent Dirichlet Allocation (LDA) approach and VADER sentiment analysis to assess emotional tone, posts on X (formerly Twitter) with hashtags #20 mph and #20 mya (Welsh language for mph) were analyzed during two periods: one-month post-implementation (September-October 2023) and six months later (February-March 2024).RESULTS: The results indicate that public opposition to the 20 mph default speed limit decreased slightly six months after its implementation but remained high. However, since opposition declined while positive sentiment increased, results suggest a positive trend, aligning with psychological theories on social influence and attitude change.CONCLUSIONS: The study underscores the value of a qualitative-quantitative approach in capturing nuanced public perspectives, offering insights beyond traditional survey methods. These findings provide actionable guidance for policymakers and practitioners seeking to implement and communicate public safety policies effectively.PRACTICAL APPLICATIONS: This research uniquely contributes to the literature by examining the 20 mph speed limit in Wales, combining computational and psychological methods to explore public opinion dynamics over time.</p
The Generalised Method of Moments and the transformation of data
No full textIt is not unusual to find empirical work for which conventional goodness of fit measures show that the conditional distribution of one set of variables on another is incompatible with the Gaussian (that is, normal) probability density. This has the important implication that conditional expectations will not, in general, be linear functions of the variables held fixed. In this paper the inverse hyperbolic sine transformation is used in conjunction with the Generalised Method of Moments (GMM) to implement asymptotically efficient parameter estimation based on the Gaussian probability density. Two examples are provided of the effectiveness of these procedures in conforming data to Gaussian distributional assumptions. The first involves the book to market ratios of equity of a large sample of publicly listed North American firms covering the period from 2005 until 2019; the second is based on an analysis of the U.S. money supply, stock prices and inflation covering the period from 1871 to 2018.</p
Mobilities, design and passenger experiences
No full textThis special issue showcases new and emerging research at the intersection of mobility studies and design, examining how transport infrastructures, environments and embodied mobile practices are shaped and ‘designed’ by a whole host of experts, professions and politicians for different purposes. In this introductory article we provide an overview of transport and mobility design, opening with a discussion of the long-standing importance of design to the history of infrastructural provision. We outline the important contribution that mobility studies approaches can make to the study of transport design, highlighting the role of qualitative and mobile methods, sociological and philosophical analyses of passenger subjectivities, and of politically sensitive approaches to the design of mobility infrastructures. We examine how passengers possess different bodily capacities and abilities which may lead them to be included or excluded, have access to services or not. In the final section we introduce the seven articles comprising the special issue.</p
Public perspectives on the 20 mph speed reduction in Wales:A Latent Dirichlet Allocation Approach
No full textINTRODUCTION: This study investigates public attitudes toward the Welsh Government's implementation of a 20 mph default speed limit on all non-unrestricted roads, for example those in urban environments, residential neighborhoods, and areas with high pedestrian activity, that previously were 30 mph, aimed at reducing traffic crashes, emissions, and improving quality of life.METHOD: Using a Latent Dirichlet Allocation (LDA) approach and VADER sentiment analysis to assess emotional tone, posts on X (formerly Twitter) with hashtags #20 mph and #20 mya (Welsh language for mph) were analyzed during two periods: one-month post-implementation (September-October 2023) and six months later (February-March 2024).RESULTS: The results indicate that public opposition to the 20 mph default speed limit decreased slightly six months after its implementation but remained high. However, since opposition declined while positive sentiment increased, results suggest a positive trend, aligning with psychological theories on social influence and attitude change.CONCLUSIONS: The study underscores the value of a qualitative-quantitative approach in capturing nuanced public perspectives, offering insights beyond traditional survey methods. These findings provide actionable guidance for policymakers and practitioners seeking to implement and communicate public safety policies effectively.PRACTICAL APPLICATIONS: This research uniquely contributes to the literature by examining the 20 mph speed limit in Wales, combining computational and psychological methods to explore public opinion dynamics over time.</p
Self-Organising Explainable Multi-View Representation Learning for Remote Sensing Scene Classification
No full textRemote sensing scene classification is widely considered to be a challenging task due to high intraclass variability and interclass similarity in remotely sensed imagery. While existing deep neural networks achieve promising performance, they often lack transparency and generalisation capability. To enhance interpretability without sacrificing accuracy, a novel self-organising transparent multi-view representation learning framework based on evolving fuzzy neural encoders for remote sensing scene classification is introduced in this paper. The framework leverages multiple pre-trained convolutional neural network backbones with different architectures to extract image embeddings from multiple views. The multi-view image embeddings are projected into a lower-dimensional feature space using multilayer evolving fuzzy neural networks trained in a supervised or self-supervised fashion as encoders and subsequently fused for scene classification. Extensive experiments on six benchmark datasets (Optimal-31, WHU-RS, UCMerced, AID, RSI-CB256, and PatternNet) demonstrate the framework’s superior performance, achieving average accuracies of 99.81 %, 98.83 %, 97.86 %, 98.37 %, 99.84 %, and 98.83 %, respectively, without fine-tuning to the specific context. Ablation studies confirm the complementary contributions of the multi-view, supervised, and self-supervised components in the proposed framework. The proposed framework provides an effective solution for remote sensing scene classification, achieving high accuracy with enhanced transparency and interpretability
Mobilities, design and passenger experiences
No full textThis special issue showcases new and emerging research at the intersection of mobility studies and design, examining how transport infrastructures, environments and embodied mobile practices are shaped and ‘designed’ by a whole host of experts, professions and politicians for different purposes. In this introductory article we provide an overview of transport and mobility design, opening with a discussion of the long-standing importance of design to the history of infrastructural provision. We outline the important contribution that mobility studies approaches can make to the study of transport design, highlighting the role of qualitative and mobile methods, sociological and philosophical analyses of passenger subjectivities, and of politically sensitive approaches to the design of mobility infrastructures. We examine how passengers possess different bodily capacities and abilities which may lead them to be included or excluded, have access to services or not. In the final section we introduce the seven articles comprising the special issue.</p
The Generalised Method of Moments and the transformation of data
No full textIt is not unusual to find empirical work for which conventional goodness of fit measures show that the conditional distribution of one set of variables on another is incompatible with the Gaussian (that is, normal) probability density. This has the important implication that conditional expectations will not, in general, be linear functions of the variables held fixed. In this paper the inverse hyperbolic sine transformation is used in conjunction with the Generalised Method of Moments (GMM) to implement asymptotically efficient parameter estimation based on the Gaussian probability density. Two examples are provided of the effectiveness of these procedures in conforming data to Gaussian distributional assumptions. The first involves the book to market ratios of equity of a large sample of publicly listed North American firms covering the period from 2005 until 2019; the second is based on an analysis of the U.S. money supply, stock prices and inflation covering the period from 1871 to 2018.</p
Fighting the next pandemic?:Civil-military collaboration in health emergencies after COVID
No full textOver the past decade, militaries have been highly visible elements in the response to health emergencies and in particular disease outbreaks. Although there has been a long tradition of civil-military collaboration in health, COVID-19 saw an unprecedented worldwide use of militaries which occurred within a permissive environment established by narratives of global health, humanitarian intervention and multi-sectorality. This creates a dilemma that militaries will likely be an important element in responding to a major health emergency, but that this risks not only militarizing health emergencies but the balance between society and the military more generally. Moreover, the response to the COVID pandemic suggested that current emergency planning is often poorly prepared for the use of militaries in health crises, thereby reducing the effectiveness of a response. This article engages with how concerns over the securitization of health have evolved into concerns over militarization, and the question of how militaries might be used effectively in future health emergencies without risking the militarization of health or damaging civil-military relations more widely
Reconceptualizing the Nation in Sanctuary Practices:Toward a Progressive, Relational National Politics?
No full textThis article explores sanctuary in Wales, focusing on the Welsh Government’s recent declaration to become a Nation of Sanctuary (NoS), and identifying how the national scale provides an alternative locus for progressive sanctuary measures. In revealing the nation’s emergence as another crucial site of sanctuary, the work reconceptualizes the nation’s place in sanctuary policies and practices in two ways: (i) it locates sanctuary through a national scale, thus moving beyond the city/state dichotomy that has dominated explanations of sanctuary, and (ii) it shows the importance of decoupling the nation-state compound while simultaneously integrating the nation(al) into discussions on sanctuary without being bound to the state or xenophobic populism. In showing how “nations against the state” can participate in sanctuary measures, we expand the current understanding of where sanctuary can be found, and capture the various forms of national belonging and identities that exist within plurinational states, including alternative, progressive forms of civic belonging. This is particularly significant in light of the tightening of state immigration policies, greater regulation of immigrant entry at state borders, and continuation of restrictive citizenship policies witnessed in recent years, which have ignited sanctuary measures aimed at creating safe spaces beyond the reach of state measures