http://hdl.handle.net/10453/35217 2026-07-01T13:27:30Z http://hdl.handle.net/10453/195458 Title: Spatial Association between Near-Misses and Accident Blackspots in Sydney, Australia: A Getis-Ord Gi∗Analysis Authors: Grigorev, A; Lillo-Trynes, D; Mihǎiţǎ, AS Abstract: Conventional road safety management is inherently reactive, relying on analysis of sparse and lagged historical crash data to identify hazardous locations, or crash blackspots. The proliferation of vehicle telematics presents an opportunity for a paradigm shift towards proactive safety, using highfrequency, high-resolution near-miss data as a leading indicator of crash risk. This paper presents a spatial-statistical framework to systematically analyze the concordance and discordance between official crash records and near-miss events within urban environment. A Getis-Ord statistic is first applied to both reported crashes and near-miss events to identify statistically significant local clusters of each type. Subsequently, Bivariate Local Moran's I assesses spatial relationships between crash counts and High-G event counts, classifying grid cells into distinct profiles: High-High (coincident risk), High-Low and Low-High. Our analysis reveals significant amount of LowCrash, High-Near-Miss clusters representing high-risk areas that remain unobservable when relying solely on historical crash data. Feature importance analysis is performed using contextual Point of Interest data to identify the different infrastructure factors that characterize difference between spatial clusters. The results provide a data-driven methodology for transport authorities to transition from a reactive to a proactive safety management strategy, allowing targeted interventions before severe crashes occur. 2025-01-01T00:00:00Z http://hdl.handle.net/10453/195416 Title: Environmental and economic impact assessment of single-use laboratory plastic waste: A case study Authors: Tan, YH; Ong, HC; Raja Ali, RA; Gew, LT Abstract: Research labs rely on disposable plastics for sterility, safety, and affordability, but their environmental and economic impact has been largely overlooked. While the impact of single-use plastics is well-known, laboratory waste poses an additional challenge due to necessary treatment processes. Hence, this study investigated both the environmental and economic impact associated with single-use plastic waste management in a university laboratory, serving as a case study for Southeast Asia. A retrospective analysis was conducted on laboratory plastic waste management between 2014 and 2023 at a Malaysian university. Three waste management methods, i.e. incineration, landfilling with microwave pre-treatment, and landfilling with ozone pre-treatment, were implemented at consecutive intervals. Environmental impact was assessed by quantifying the GHG emissions in terms of carbon dioxide equivalent (CO₂eq) emissions, while economic assessment was evaluated based on invoiced costs. A total of 29,180.11 kg of single-use plastic waste resulted in 46,420.08 kg CO₂eq emissions and a cumulative disposal cost of RM 84,890.49. Among the three methods, landfilling with ozone pre-treatment demonstrated the lowest environmental impact (1.55 kg CO₂eq/kg) and cost (RM 2.79/kg). Direct emission from the end-of-life stage was the main source of GHG emissions, while disposal fees represented the largest portion of total costs. In light of these findings, it is crucial for universities and research institutions to recognise the environmental and economic impact of laboratory single-use plastics. This case study will serve as a foundation for advancing pre-treatment technologies and future end-of-life solutions. 2025-12-01T00:00:00Z http://hdl.handle.net/10453/195386 Title: Threat Detection in Ethereum Smart Contracts Using a Hierarchical Depthwise Graph Convolutional Neural Network Authors: Le, TH; Pham, T; Nguyen, DN; Dinh, H; Le, TH 2026-01-01T00:00:00Z http://hdl.handle.net/10453/195351 Title: Enhanced Density Peak Clustering for High-Dimensional Data Authors: Wang, Z; Yang, J; Guan, J; Zhang, C; Liang, X; Jiang, B; Sheng, W Abstract: As a foundational clustering paradigm, Density Peak Clustering (DPC) partitions samples into clusters based on their density peaks, garnering widespread attention. However, traditional DPC methods usually focus on high-density regions, neglecting representative peaks in relatively low-density areas, particularly in datasets with varying densities and multiple peaks. Moreover, existing DPC variants struggle to identify clusters correctly in high-dimensional spaces due to the indistinct distance differences among samples and sparse data distributions. Additionally, existing methods typically adopt a one-step label assignment strategy, making them prone to cascading errors when initial misassignments occur. To address these challenges, we propose an Enhanced Density Peak Clustering (EDPC) method, which creatively incorporates multilayer perceptron (MLP)-based dimensionality reduction and a hierarchical label assignment strategy to significantly improve clustering performance in high-dimensional scenarios. Specifically, we introduce an effective selection condition that combines average densities and density-related distances to generate potential cluster centers, ensuring that peaks across different density regions are considered simultaneously. Furthermore, an MLP, guided by pseudo-labels from sub-clusters, is designed to learn low-dimensional embeddings for high-dimensional data, preserving data locality while enhancing clusterability. Extensive experiments demonstrate the effectiveness and superiority of EDPC against state-of-the-art DPC methods. 2025-04-11T00:00:00Z