马来西亚工艺大学

Distracted Driver Detection (DDD) is critical for intelligent transportation systems, yet existing methods often incur high computational costs, leading to non-real-time detection on terminals and/or limited generalization across different camera views. In this work, we propose D4RNet, a lightweight Distracted Driver Detection network that integrates knowledge Distillation with a gradient Reversal layer (GRL) for the first time in this domain. We employ a Swin-B as a teacher guide a MobileNetV2 student for training, in addition to predicting the type of distraction, the forward path extracts multi-scale, low-semantic features to predict the auxiliary tags, while during backpropagation the GRL reverses the gradient to adversarially suppress the auxiliary label, thereby enforcing the model to learn domain-invariant representations of the distraction behaviors, rather than noise such as image modality, camera angle, or vehicle scenes. Experimental results show improvements of 3.4% on D-all and 4.3% on N-all over the baseline, attains 81.32% average accuracy over eight subsets, surpassing the previous SOTA by 3.36%, while achieving 52 FPS on a real edge device, Jetson Nano via ONNX Runtime. Qualitative tests on real driving videos also demonstrate strong performance and robustness. Moreover, we introduce five new temporal–spatial combined splits for the 100-Driver dataset along with corresponding baselines, to provide a more comprehensive evaluation of model performance. We also indicate, based on numerous experiments, that future work should prioritize annotation quality and image diversity rather than architectural over-engineering. In summary, D4RNet offers a fast, accurate, and open-source solution for real-time DDD at the edge. Code, data, and models are available at https://github.com/jerodzhao/D4RNet.

Jacking force prediction for pipe jacking in highly weathered rock is challenging because closed-form mechanistic models require reliable strength parameters that may be unavailable when intact cores cannot be recovered. Using two full-scale microtunnelling drives in weathered phyllite (120 m) and sandstone (140 m), this study develops an interpretable deep learning framework to relate operational parameters to drive-specific jacking force variability. Before developing the attention-based interpretation framework, GRU, LSTM and Conv1D baseline models were first compared using separate random grid search hyperparameter optimisation, with validation data derived only from the training set and the original testing data retained as unseen records. The selected GRU model was then integrated with Bahdanau attention for one-step-ahead prediction. To prevent temporal information leakage, the section-wise train-test split was completed before training-only up-sampling and sliding-window reconstruction, and no window was allowed to cross the train-test boundary. The GRU model achieved the best overall performance, with R2 of 0.82 and RMSE of 7.93 ton for phyllite, and R2 of 0.97 and RMSE of 12.22 ton for sandstone. Spatial attention was used to identify operational parameters with higher predictive contribution along the drive, while temporal attention identified when each parameter became predictively influential. These outputs were combined through a post-hoc dot-product mapping to produce a spatial–temporal influence score. Cross-drive transfer learning showed poor direct transfer in both directions, although predicted-drive fine-tuning improved performance, indicating limited direct generalisability but improved adaptability after fine-tuning. As the two drives also differ in burial depth, pipe diameter and machine configuration, the attention results are interpreted as drive-specific model-attributed patterns rather than lithology-only effects. The phyllite model relied more strongly on face support and thrust response, whereas the sandstone model showed greater predictive contribution from drive length, elapsed time and jacking speed. The resulting spatial–temporal influence map provides a practical framework for interpreting operational variability and stage-dependent predictive sensitivity during pipe jacking under changing geological and project-specific conditions.

Marine traffic safety and efficiency are critical concerns, especially with the increasing complexity of shipping environments. While automation holds promise for improving these aspects, autonomous ships’ ability to navigate complex areas remains uncertain. The intricate characteristics of marine traffic, including emergency behaviour, harsh environmental conditions, traffic crossing situations, and the potential for human error on traditional ships, pose significant challenges for deploying autonomous ships. Therefore, this study explores the potential Digital Twin (DT) applications in marine traffic environments, aiming to enable safe and efficient Autonomous Ship (AS) navigation. The current rules, navigation, and path-following systems of AS are reviewed to assess their competency. At the same time, macroscopic traffic analysis models are outlined to assess the possibilities of extending navigation systems for operating in complex areas. DT is an emerging technology in the maritime industry, driven by elements of Industry 4.0 such as Artificial Intelligence (AI), machine learning, and big data. It can potentially accelerate the development of AS through the integration of macroscopic traffic information, enhancing operational safety and efficiency, as well as providing decarbonisation opportunities in the marine environment. Hence, DT is likely to facilitate a highly automated and environmentally sustainable maritime transport network, contributing to the realization of smart mobility.

Abstract This study examines the role of waqf as a socio-economic institution, highlighting its contributions to social welfare, poverty alleviation, and sustainable development. By addressing governance inefficiencies and operational constraints, the paper explores how waqf can serve as a tool for economic empowerment and equitable resource distribution in contemporary societies. A multi-method review was conducted following PRISMA guidelines, analyzing 180 peer-reviewed articles from Scopus. Using Structural Topic Modeling (STM), a machine learning approach, the study identifies key themes in waqf management, governance, and policy integration. The analysis reveals ten key themes, including waqf development models, integration with Islamic social finance, governance frameworks, and its role in education, healthcare, and social protection. The findings highlight persistent challenges, including legal ambiguities, institutional inefficiencies, and underutilized assets, which hinder waqf’s full potential in addressing socio-economic disparities. Strengthening waqf governance, implementing policy reforms, and integrating modern financial and administrative frameworks can enhance its role as a sustainable economic institution. This study uniquely contributes to social economic discourse by applying machine learning-based analytical methods to examine waqf’s evolving role in sustainable development and social policy. The findings offer insights for policymakers, researchers, and institutions on optimizing waqf as a transformative tool for welfare and economic resilience.

Tsunamis are among the most destructive natural phenomena instead of disasters, particularly for coastal communities that have low levels of socio-economic resilience. The research this article is based on critically evaluate and synthesize the socio-economic indicators used in tsunami vulnerability assessments through a systematic review approach. The findings indicate that vulnerability indicators can be classified into four main domains: exposure, early warning capacity, evacuation and emergency response capacity, and recovery capacity. However, there is inconsistency in the selection of indicators, a lack of adaptation to local contexts, and an excessive reliance on physical factors rather than social dimensions. This article contributes to the development of a conceptual framework that integrates socio-economic indicators into tsunami vulnerability modelling, particularly for developing countries that are underrepresented in the global literature. The results provide an important scientific basis to support the development of vulnerability models that are more local, community-based, and responsive to varying levels of social inequality. The implications of this research are in line with the aspirations of the Sustainable Development Goals (SDGs), particularly SDG 11: Sustainable Cities and Communities and SDG 13: Climate Change Action, through recommendations for improvements to disaster response plans and community-based risk mapping. The research also suggests that future research should emphasize the integration of socio-economic indicators with geospatial technologies and artificial intelligence (AI) approaches to produce dynamic, adaptive, and community-oriented tsunami risk mapping.