85000 research outputs found
Sort by
Comprehensive Evaluation of Bankruptcy Prediction in Taiwanese Firms Using Multiple Machine Learning Models
No full textBankruptcy prediction is a significant issue in finance because accurate predictions would enable stakeholders to act quickly to reduce their financial losses. This study developed an advanced bankruptcy prediction model using Support Vector Machines (SVM), Random Forest (RF), and Artificial Neural Network (ANN) algorithms based on datasets from the UCI machine learning repository. The core contribution of this research is the establishment of a hybrid model that effectively combines multiple machine learning (ML) algorithms with advanced data with the Synthetic minority oversampling technique Tomek (SMOTE Tomek) or SMOTE- Edited Nearest Neighbor (SMOTE-ENN) resampling data technique to improve bankruptcy prediction accuracy. Additionally, a wrapper-based feature selection (FS) utilizing Binary Particle Swarm Optimization (BPSO) was utilized to find an optimal feature subset and boost the model’s predictive performance. After selecting the best features, these were used to train the three ML algorithms, and hyper-parameter optimization was implemented to boost model performance. From the results measured by evaluation metrics, the proposed model ANN with the combination of parameter tuning, feature selection algorithm, SMOTE-ENN, and optimal hyper-parameters demonstrates superior performance compared to traditional methods, achieving an F1 Score of 98.5% and an accuracy of 98.6%. The results suggest that the predictive performance of bankruptcy models can be significantly enhanced by integrating multiple analytical methodologies. This approach not only improves the accuracy but also the reliability of financial risk assessments, providing valuable insights for investors, financial analysts, and policymakers. The success of the model opens avenues for further research into hybrid predictive models in various sectors of finance, potentially transforming risk assessment methodologies.</p
Minecraft WoZ Study Dataset
No full textPlease fill out this terms of use form before accessing the dataset: https://forms.gle/N8gk6NNsJaBW9AKE7This is the accompanying data published alongside a manuscript detailing an experiment that involved an adaptive companion character through real-time physiological measurement of cognitive load, who was controlled by the wizard, the primary researcher.This dataset presents an array of physiological and log data collected during Minecraft gameplay. It contains the raw EMG, PPG, EDA, game logs, and screen-recorded gameplay from 16 participants. It also contains the scripts for processing the PPG, EDA, and EMG data in real-time, as well as the task instructions and hints provided by the companion. Full details of the data collection (including hardware and software specifications) can be found in the accompanying publication "Real-Time Adaptation of a Non-Player Character Companion Using Physiological Signals of the Player" by E J Pretty, R Guarese, J Hamari, H M Fayek and F Zambetta.Variable DescriptionsEMG: Measured in volts (V), collected from Myoware Muscle Sensor 1.0EDA: Measured in microsiemens (µS) by the EmotibitPPG: Raw arbitrary units measured by the EmotibitGameplay: Captured by OBS</p
Samuel L. Jackson in Geriaction: “Bad Ass” Black Screen Masculinity, Aging, and Redundancy
No full textDonnar examines how Samuel L. Jackson’s roles in geriaction—Hollywood action films featuring aging stars—interrogate the all-too-readily presumed whiteness of the cycle. For Donnar, Racquel Gates’ 2018 concept of the “double negative” explains scholarly underappreciation of Jackson as an action star, allowing us to recognize the positive, disruptive potential of his post-2000s Black action masculinity. Jackson’s articulation of action stardom avoids the emphasis on muscularity, bodily display, wisecracking, and Everyman ingenuity that characterizes most scholarship on geriaction, and that masks decreases in physical strength and agility. Jackson’s leading, co-starring, and supporting performances embody a different model of geriaction stardom and permit fuller consideration of biracial buddy pairings, intergenerational dynamics, and the place of villainy.</p
Effect of construction defects on construction and demolition waste management in building construction: a systematic literature review
No full textAbstract
The occurrence of defects in building construction projects is a significant issue, leading to increased construction waste and negatively affecting sustainability and overall project performance. Despite its critical nature, the specific relationship between construction defects and waste generation has been underexplored in the literature. This study seeks to address this gap by conducting a systematic literature review of relevant publications. The research followed the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines, conducting an extensive search across databases like Scopus and Web of Science, which resulted in the identification and content analysis of 59 pertinent articles. The findings reveal that poor workmanship, inadequate planning and scheduling, and frequent design changes are the primary causes of defect-related waste. Additionally, the study identified 12 themes, noting that the quantification of the cost of quality and the association between defect, rework and waste have not been thoroughly analyzed. The study's implications are twofold: Theoretically, it contributes to the academic understanding of the link between construction defects and waste generation, laying a foundation for future research in this area. Practically, it underscores the need for improved industry practices, such as enhanced training for construction workers, more rigorous project planning, and stricter adherence to design and specifications, to mitigate defect-related waste and promote sustainable construction practices.</p
Monte Carlo Simulations of Space Radiation Exposure for Astronauts in Spacecraft
No full textThe imminent expansion of crewed space exploration, particularly towards lunar missions, necessitates a comprehensive understanding of the inherent risks posed to humans by the space environment. Among many technological challenges and risks, the threat of ionising radiation from Galactic Cosmic Rays and Solar Energetic Particles poses significant health risks during extended missions beyond Earth’s protective magnetosphere. High energy particles interact with spacecraft structures, undergoing scattering, spallation and fragmentation (among other interaction mechanisms), generating a myriad of secondary particles, triggering a particle shower. Any of these particles may traverse human tissue inducing complex biological responses that increase the likelihood of cancer morbidity and mortality, as well as adverse cognitive or behavioural conditions of the crew over their lifetime. Advanced radiation mitigation strategies and accurate risk assessment methodologies are required to safeguard the health and well-being of crew members during extended missions to lunar orbit and beyond. Due to the difficulty in replicating the space radiation environment for groundbased experiments, particle transport simulations through spacecraft materials are important for evaluating and investigating risk mitigation strategies for specific mission scenarios. The scientific objective of this work is to develop a validated application that utilises the Geant4 toolkit to simulate 3D particle transport of the space radiation environment in spacecraft in a timely and efficient manner. The Geant4-based application developed throughout the project provides an easy-to-use platform to solve space radiation problems, and is named ’ASPIRE’ (Assessment of SPacecraft Ionising Radiation Environment). A novel methodology was employed, integrating state-of-the-art particle flux models (BON2020, AP8), updated Geant4 physics models and cross-sectional data, and ICRP-123 recommendations for scoring, to estimate absorbed dose and dose equivalence in human tissue and organs. Four 3D 1:1 scale, medium fidelity geometric models of ISS configurations 10A, 11A, 20A, and Spx-20 were developed to be used as representative spacecraft. The functionality of the application to modify the spacecraft geometry, including various passive shielding concepts, was a key factor in its development incorporating imports from external software such as Solidworks™. Monte Carlo simulations of the space radiation environment are inherently computationally intensive due to factors such as the high energy of primary particles, the large number of interactions and secondary particles generated, and the considerable size of spacecraft structures relative to the points of interest. A novel variance reduction technique is applied to reduce computation time of the space radiation simulations while maintaining scientific plausibility. This is achieved by reducing the effective isotropic region around the scoring volume. A reduction of the isotropic region radius from 100 m to 20 m produced an efficiency improvement of 62% and a reduction in computation time by 38%. A reduction to 10 m produced a 300% increase in efficiency and reduced computation time by 75%, however a smaller isotropic radius will underestimate dose. ASPIRE validation and benchmarking was conducted against the MATROSHKA anthropomorphic human phantom experiments (MTR-1, MTR-2A, MTR-2B and MTR-KIBO) on the International Space Station between 2004 and 2011. Simulated absorbed dose rates in the eye lens, skin, kidneys, small intestine, lungs, and stomach are compared to thermoluminescent detectors from ’organ boxes’ placed in approximate locations of the organs. The MTR-1 organ absorbed dose rates vary between -15.2% and 3.2%, MTR-2A between -15.2% and -26.5%, MTR-2B between -14.3% to - 17.6% and MTR-KIBO between 15% to 38.8%. The simulated dose distribution throughout the organs of the human body showed good agreement with experimental results. For lightly shielded astronauts (eg. MTR-1), the skin, salivary gland, breast and eye lens had a high dose rate, driven largely by trapped protons. The inner organs had much lower dose rates, demonstrating the self-shielding of the human body. As the shielding increased, the GCR became the dominant dose rate contributor, and produced a more uniform dose rate amongst the organs. Similar distributions from light and heavy shielded modules are seen in experimental data. The model also successfully reproduces the same trend in experimental results when comparing extravehicular to intravehicular dose rates. Causes of the systematic error have been explored and several have been identified. Those that remain elusive are left as recommendations for future research. To demonstrate a typical use cases of ASPIRE, the boundary conditions were modified in the MATROSHKA simulations to estimate the change in absorbed dose rate, and dose equivalent rates in Lunar Orbit (at 1AU). The objective is to estimate the radiation dose to astronauts on board the Lunar Gateway. It is assumed that the International Space Station resembles the typical size, shape, and materials of the Gateway, and therefore, the validated MATROSHKA ISS geometries and materials were used in the simulations. The simulation results indicate that organ dose equivalent from GCR exposure may increase by approximately 60 to 90% between solar maximum and solar minimum conditions. In comparison to LEO MATROSHKA simulations, there is a 11% to 105% increase in whole body effective dose equivalent for MTR-1, 96% to 217% increase for the lightly shielding MTR-2A, 115% to 246% increase for the heavily shielded MTR-2B and a 75% to 203% increase for MTR-KIBO. Thus, the work provides estimates of how much higher the doses would be in a similar spacecraft placed in a location that is exposed to Galactic Cosmic Rays and not benefiting from the shielding of Earth’s magnetosphere. Furthermore, the software tool developed provides the means for comparing different structural approaches to inhabited modules and for estimating the effectiveness of new shielding strategies under development.</p
Descriptor-driven design of carbon nitride for visible light photocatalysis
No full textPhotocatalysis utilizing carbon nitride (CN) based photocatalysts presents an eco-friendly solution to energy challenges. Despite progress in enhancing CN performance, targeted design for specific applications remains challenging due to the complex feature-activity relationships. A computation-assisted strategy is proposed to explore multidimensional correlations between electronic properties and photoactivity in CNs for various applications, identifying d/p-band centers and effective mass as key descriptors for CN photocatalyst design. Specifically, the d-band center of the co-catalyst (Pt) correlates with H* dissociation energy, serving as a descriptor for designing hydrogen evolution reaction (HER) photocatalysts: the N–C p-band center difference, closely linking to O2 adsorption and activation, emerges as a valuable indicator for H2O2 generation. These descriptors guide CN photocatalyst design through defect engineering, leading to a 6.7-fold increase in HER and 24.1-fold boost in H2O2 generation compared to pristine CN. Mechanistic analyses further reveal deeper structure-performance relationships, illustrating the influence of CN local structure on the stability of critical intermediates and the energy barriers of rate-limiting steps. By integrating computational and experimental methods, this study establishes a robust framework for the rational design of CN-based photocatalysts. This approach has significant potential for extension to other photocatalytic systems, offering broader applications in energy and environmental fields.</p
Characterising the Nanostructure of Lyotropic Liquid Crystal Phases Mediated by Protic Ionic Liquids
No full textLyotropic liquid crystal phases (LCPs) arising from the self-assembly of lipids represent a complex and disordered class of bulk materials. The ability of lipids to self-assemble into distinct LCPs depends upon solvent interactions, solvent type, solvent concentration, and temperature. Due to this, monoolein (MO)- based LCPs have been studied for biological applications including pharmaceuticals and drug delivery. Protic ionic liquids (PILs) have demonstrated versatility as solvents, particularly in mediating the selfassembly of amphiphiles. While individual contributions of PILs, MO, and LCPs have been studied in fields like protein crystallisation, a comprehensive understanding of LCPs within lipid:PIL systems is lacking. Furthermore, lipid:PIL LCPs are scientifically relevant media suitable for testing recent advancements and techniques. The most widely used method for structure determination of LCPs is small-angle X-ray scattering (SAXS). Through SAXS, information about phase stability, phase transitions and lattice parameters can be obtained. However, the inherent random orientational order of LCPs poses challenges in extracting three-dimensional structure information through conventional methods like SAXS. Utilizing advanced X-ray scattering techniques, a promising technique for detailing nanostructure has emerged. Scanning a microfocused beam across a sample allows fluctuation X-ray scattering (FXS) patterns to be gathered with statistical sensitivity to the local three-dimensional structure. The nanostructure of LCPs can be investigated by analysing the data with the recently developed pair angle distribution function (PADF). This thesis begins with a study of the thermal stability of PILs to determine if they are suitable candidates for mediating MO-based LCPs. The thermal stability of six PILs, namely ethylammonium nitrate (EAN), ethanolammonium nitrate (EtAN), ethylammonium formate (EAF), ethanolammonium formate (EtAF), ethylammonium acetate (EAA), and ethanolammonium acetate (EtAA), were investigated by subjecting samples to 60 °C heating for either one hour or one week, under both sealed and open conditions. Changes in mass, pH, water content, thermal phase transitions and molecular structure were assessed. In PILs heated for one hour and one week in sealed containers, as well as those heated for one hour in open containers, negligible change to physiochemical properties or molecular structure was evident. After one week exposure in open containers the physiochemical properties and thermal transitions of EAN and EtAN varied, while their molecular structure remained stable. Contrary to the nitrate based PILs, EAA and EtAA showed instability in their thermal transitions, while their physiochemical properties remained unchanged. EAF and EtAF displayed the greatest instability of the 6 PILs chosen for the study, with NMR spectra of both formate based PILs displaying peak coupling indicative of formamide formation. Despite the structural changes after one week, the negligible shifts in short time frames, and stability in sealed containers, determined that all 6 PILs were suitable for use in LCP investigations. LCPs of MO in each of the six PILs were then investigated to examine the role of PIL chemical structure in directing amphiphile self-assembly. Specifically, this study detailed how increasing the alkyl chain length of the cation or anion, the presence of a hydroxyl group in the cation, or varying the anion, affected phase formation. Using SAXS and cross polarised optical microscopy (CPOM), the LCPs of MO in the 6 PILs between 20–80 wt.% MO were characterized over a temperature range from 25 to 65 °C. It was found that PILs containing ethylammonium cations promoted lamellar and bicontinuous cubic phases. Ethanolammonium based PILs were found to support inverse hexagonal and bicontinuous cubic phases. Formate and acetate anions broadened the stability range of the bicontinuous cubic phase when compared to the nitrate anions. As a precursor to real-space structural analysis of MO-PIL LCPs, spatial mapping, FXS techniques and angular intensity correlation calculations were developed and applied to a 75 wt% MO 25 wt% water system supporting a gyroid cubic (Ia3d) cubic phase. MO-based Ia3d phases have been studied for applications in drug delivery and biological systems, yet real-space nanostructural analysis of an Ia3d phase has not previously been performed with FXS. Modifying an X-ray fluorescence microscopy (XFM) beamline, microfocused SAXS patterns were obtained. Spatial mapping techniques were developed which identified homogenous Ia3d domains of the sample, then subsequent angular intensity correlation calculated using data from these domains. This allowed selective extraction of diffraction patterns with consistent structural features, an important prerequisite for real-space analysis with the PADF. The spatial mapping and angular intensity correlation analysis revealed heterogeneity in domain size and orientation that standard SAXS could not resolve. Furthermore, the methods demonstrated the necessity of data filtering prior to angular correlation and PADF calculations. This study established the methodological pipeline for real-space structural analysis, which was subsequently applied to MO–PIL systems. Furthermore, this investigation enabled a direct comparison between similar phases seen in MO-water and MO-PILs. A detailed spatial and fluctuation scattering analysis was performed on a selected MO–PIL system composed of 60 wt% MO and 40 wt% EAF. At this composition, MO-EAF had been identified as forming an Ia3d cubic phase between 30 and 45 °C. The real-space structure of the Ia3d phase was examined by calculating PADFs from the angular correlations, revealing characteristic features suggestive of progressive structural disordering that could not be identified by SAXS analysis alone. To summarise, in this thesis I develop a comprehensive picture of PIL-lipid interactions. The methods and insights developed within this thesis include evaluating the thermal stability of PILs, phase behaviour characterisation, microfocused spatial mapping, angular intensity correlation analysis and PADF real-space interpretation of LCPs. The presented work establishes novel tools for understanding the nanostructure of disordered materials, providing pathways for targeted solvent and nanostructure characterisation based on real-space investigations.</p
A Novel Blockchain-Enabled Federated Learning Scheme for IoT Anomaly Detection
No full textIn this research, we proposed a novel anomaly detection system (ADS) that integrates federated learning (FL) with blockchain for resource-constrained IoT. The proposed system allows IoT devices to exchange machine learning (ML) models through a permissioned blockchain, enabling trustworthy collaborative learning through model sharing. To avoid single-point failure, any device can be a centre of the FL process. To deal with the issue of resource constraints in IoT devices and the model poisoning problem in FL, we introduced a novel method to use commitment coefficients and ML model discrepancies when selecting particular devices to join the FL process. We also proposed an efficient heuristic method to aggregate a federated model from a list of ML models trained locally on the selected devices, which helps to improve the federated model’s anomaly detection ability. The experiment results with the popular N-BaIoT dataset for IoT botnet attack detection show that the proposed system is more effective in detecting anomalies and resisting poisoning attacks than the two baselines (FedProx and FedAvg).</p
Hyperelastic modelling of tyre barrier for racetrack crash safety
No full textTyre barriers offer a sustainable and cost-effective solution for improving roadside safety by repurposing end-of-life tyres into energy-absorbing crash mitigation systems. However, accurately modelling their complex multilayer hyperelastic structure remains challenging due to the nonlinear behaviour of rubber materials and the high computational cost associated with detailed finite element (FE) models. This paper presents a validated and computationally efficient modelling approach for simulating tyre-based safety barriers using only shell and beam elements. A single-row tyre barrier comprising uninflated tyres, bolted connections, and preloaded straps was selected as the case study. The developed FE model was validated against full-scale crash test data, demonstrating high correlation in impact response, acceleration, and energy absorption. The study further investigates the effects of strap preload, conveyor belt attachments, and polymer tube inserts on barrier performance. Results suggest that accurate modelling of preload significantly improves simulation fidelity, and that structural enhancements notably increase impact energy absorption. The proposed method provides a robust and efficient tool for engineers and policymakers to design, evaluate, and optimise tyre-based roadside safety barriers for different road crash scenarios.</p