Heriot-Watt University Malaysia

We study adversarial learning when the target distribution factorizes according to a known Bayesian network. For interpolative divergences, including $(f,Γ)$-divergences, we prove a new infimal subadditivity principle showing that, under suitable conditions, a global variational discrepancy is controlled by an average of family-level discrepancies aligned with the graph. In an additive regime, the surrogate is exact. This closes a theoretical gap in the literature; existing subadditivity results justify graph-informed adversarial learning for classical discrepancies, but not for interpolative divergences, where the usual factorization argument breaks down. In turn, we provide a justification for replacing a standard, graph-agnostic GAN with a monolithic discriminator by a graph-informed GAN (GiGAN) with localized family-level discriminators, without requiring the optimizer itself to factorize according to the graph. We also obtain parallel results for integral probability metrics and proximal optimal transport divergences, identify natural discriminator classes for which the theory applies, and present experiments showing improved stability and structural recovery relative to graph-agnostic baselines.

With increasing pressures from climate change and urban expansion, the development of resilient “sponge cities” is essential to mitigate flooding and reduce pollution. Rain gardens represent a key green infrastructure intervention and have the potential to be implemented far more widely in new developments or retrofitted into existing ones. Rain gardens are particularly appealing to urban planners because they can deliver multiple co-benefits by enhancing biodiversity and amenity while achieving water management objectives. However, gaps in the understanding of rain garden hydrology remain a barrier to widespread adoption. In contrast to grey infrastructure, which is supported by extensive empirical research, confidence in the hydraulic performance of vegetated systems remains limited. To embed rain gardens more effectively in urban design, their hydrological functioning must be quantified more accurately and design parameters refined. A major source of uncertainty lies in the behaviour of rooted soils. Recent studies highlight that root-oriented preferential flow can substantially increase soil hydraulic conductivity, reduce surface runoff and prevent sediment from clogging drainage structures. Plant roots may also improve soil water retention, enhance rainfall interception, attenuate peak flow and support pollutant removal. Yet despite this growing awareness, these mechanisms remain poorly quantified and are rarely represented in models of green infrastructure. As a result, current engineering design typically relies only on physical soil parameters, without accounting for dynamic plant–soil interactions. This study investigates the influence of root-oriented preferential flow on rain garden hydrology through a mixed-methods approach combining laboratory experimentation, field observation and mathematical modelling. The first phase involves single-plant mesocosms in a three-year longitudinal laboratory study of rooted soil hydrology, complemented by regular MRI imaging to capture root architecture development. This study presents initial findings from this longitudinal experiment, demonstrating how high-resolution MRI scanning can be integrated with continuous hydrological monitoring to reveal emerging flow pathways in rooted soils. These data will inform a mechanistic model that quantifies the effects of preferential flow across different root types and depths, providing new parameterisations for use in rain garden performance models.

This research explores how international students in Malaysian higher education develop digital literacy, the barriers that impede such development, and strategies that may enhance institutional support. To understand their experiences using digital tools within an academic context, semistructured interviews were conducted with international students in a qualitative design. The results showed that confidence develops through repeated use and structured support, and barriers include complexities of software applications, integration challenges across different platforms, and limited chances for hands-on training. Participants called for targeted workshops, peer mentorship, improved resource access, and early integration into the curriculum. This paper offers practical suggestions useful to universities seeking to foster international students' adjustment to digitally mediated learning environments.

Constitutive models for anisotropic clays incorporate a tensor-valued quantity named the ‘fabric tensor’ to describe the direction-dependent mechanical response of the material. Although the term ‘fabric’ reflects the directional properties at the microscale, this tensor is not actually measured or observed at the microscale but formulated as a mathematical entity and calibrated by best-fitting experimental observations at the macro-scale. This paper presents a first attempt to bridge the gap between micro and macro scales by using direct measurements of the fabric tensor at the microscale to inform a continuum-based constitutive model. Owing to the scarcity of experimental measurements of the fabric in clayey geomaterials, this paper turns to virtual experiments using the discrete-element method (DEM) to quantify the microstructural arrangement and its evolution in response to imposed stress or strain history. The virtual experimental programme was performed in a simplified two-dimensional numerical framework and consisted of a set of virgin radial paths to generate different macroscopic anisotropic responses, quantified by way of the elastic stiffness in the horizontal and vertical direction. An existing constitutive model developed within the framework of thermodynamics with internal variables (TIV) was then used to describe the macroscopic behaviour of the DEM specimens, once the fabric-related parameters had been inferred from particle orientations. The DEM-based TIV model was proven to simulate satisfactorily the numerical compressibility curves for radial compression paths at different stress ratios and, most importantly, to reproduce well the macroscopic anisotropic elastic stiffness and its evolution.

Purpose This paper explores the practical implications of community ownership in Scottish professional football, examining how governance, management and strategic priorities evolve under supporter control. Design/methodology/approach The study draws on qualitative semi-structured interviews with key actors from SPFL clubs under supporter ownership. Data were analysed using thematic analysis, with overlapping themes refined into three integrated areas. Findings Community-owned clubs prioritise financial prudence, transparency and community engagement, reframing football success as a by-product of stability and social value. However, persistent challenges include governance conflicts, volunteer fatigue and tensions between supporter expectations and professional management. Originality/value The paper contributes to debates on alternative ownership models by providing rare empirical evidence on post-transition governance in supporter-owned clubs. It highlights the conditions that enable successful transitions, the fragility of the model and lessons for policy, practice and the wider football ecosystem.

Community-based energy initiatives, such as local energy markets (LEMs), are gaining attention as a means to foster a more equitable and participatory energy transition. To fulfill this potential, local communities require robust energy-sharing mechanisms that ensure the fair distribution of financial benefits among participants. This paper presents a novel algorithm for virtual energy distribution in a Peer-to-Community (P2C) local energy market. The algorithm first determines the optimal number of participants in each time period of the LEM, based on their current energy supplier tariffs, to ensure benefits for all participants. It then applies an egalitarian energy-distribution method, called the Glass-Filling algorithm, to allocate locally produced energy within the community. Using real-world data from a UK trial involving 200 households, half of which are prosumers with solar PV and residential batteries, we demonstrate that the proposed mechanism improves equity in energy sharing compared with a benchmark double-auction mechanism. The proposed market mechanism yields an average annual bill reduction of 9.9% for producers and 5% for consumers.

The impact of generative artificial intelligence (GenAI) on higher education has been widely discussed since the public release of ChatGPT-3.5 in late 2022. However, there has been little empirical research on changes in English-for-Academic-Purposes (EAP) assessment practices in response to GenAI. This qualitative case study intends to fill this gap by examining how Scottish universities changed EAP assessments in response to GenAI, how effective those changes were perceived by EAP academics, and what recommendations EAP academics offered for future assessment practices. Data were collected from six semi-structured interviews conducted with EAP academics at five Scottish universities in mid-2024 and thematically analysed. The findings reveal that while substantial changes in assessment task design were limited, modifications to task requirements (e.g., GenAI declarations, context-specific prompts) and grading practices were more common. Moreover, our participants expressed scepticism about the effectiveness of some changes (e.g., AI use declarations) but positively perceived others (e.g., the use of context-specific questions, spontaneous speaking tasks, and named marking). As for their recommendations, the participating EAP academics generally advocated authentic and innovative tasks, such as portfolio-based assessment, reflections, multimodal projects, and GenAI output evaluation over reverting to traditional exams while simultaneously highlighting issues with workload and learning outcomes. The study implies a need for clearer institutional guidance, ongoing professional dialogue, and support for experimentation with GenAI-integrated assessment design in EAP contexts. • Substantial changes in EAP assessment task design were limited. • Modifications to task-specific requirements and grading were more common. • Participants were skeptical about procedural changes like AI use declarations. • They were positive about using specific and spontaneous tasks and named marking. • Participants advocated innovative tasks over reverting to traditional exams.

Maritime traffic poses a variety of risks to both the marine environment and marine wildlife. To quantify and predict risk, accurate data on the distribution and densities of vessel traffic is required, yet currently there is no single data type that captures all vessel traffic. Most commonly, AIS (Automatic Identification System) vessel tracking data is used, despite awareness that AIS data does not fully capture all vessels present. Therefore, evaluations using only AIS likely underestimate the potential impacts. To estimate the scale of underestimation, vessel presence within six of Scotland's Marine Protected Areas (MPAs) were recorded during >1800 h of land-based and at-sea surveys, and compared with AIS data collected from a network of receivers deployed around Scotland. Non-AIS vessels were present within MPAs during 62 % of the surveyed period, with 64 % of vessels sighted not broadcasting AIS. AIS transmission rates varied between MPA, season and vessel type. Given that AIS data is the most commonly used data type for quantifying vessel activity and predicting associated impacts, consideration must be given to the volume of vessel traffic not represented within AIS datasets, particularly within MPAs. Underestimation of actual vessel traffic is likely leading to insufficient management or mitigation efforts within areas designated for protection. • AIS (Automatic Identification System) data often used to represent vessel traffic. • Surveys conducted across Scottish Marine Protected Areas. • AIS data did not represent vessel traffic within MPAs 62 % of the time. • 64 % of powered vessels sighted within MPAs were not broadcasting AIS. • Risk predictions using only AIS would underestimate the scale of potential impacts.