RSS 1.0Computer, Engineering and Natural Sciences
Sub-communities within this community
Collections in this community
Recent Submissions
-
Fish and coral communities shape elasmobranch reef use in southern MozambiqueFew studies have explored whether a reef’s bioecological structure affects the presence of elasmobranchs. To examine if the structure of a reef’s fish and coral community influences the likelihood of observing certain elasmobranch species, we deployed a remote underwater video station (RUVS) on four reefs in southern Mozambique. A single RUVS was deployed monthly on each reef for 12 months, resulting in 48 deployments and 140 h of video recordings. Images were extracted from the video recordings to estimate the relative abundance of teleost fish and following each camera deployment a 30 m2 belt transect was completed to measure the percentage cover of corals. Coral and fish abundances were then separated into common functional metrics describing each community. NMDS and PERMANOVA were used to estimate if the calculated metrics and observations of elasmobranchs by RUVS varied between the four reefs. Metrics were then analysed for their influence on the composition of each reef’s elasmobranch community within the NMDS ordination space. The relative abundance of coral species was primarily found to be linked with the depth of the reef surveyed. Relative abundances of coral measured on the shallow reef site were distinct from the other examined reefs in ordination space, with less coral cover and a lower overall abundance of teleost and elasmobranch fish. The richness and abundance of teleost fish species, particularly piscivorous fish, was highest on the northern reef where the elasmobranch community was dominated by several species of reef shark. The southern reef also had a distinct richness and abundance of teleost fish species, with a heightened abundance of herbivorous and cleaner fish, and the observed elasmobranch community was mostly comprised of Mobula rays and guitarfish. Our findings suggest that fish and coral communities can significantly differ between reefs with similar abiotic conditions in a relatively small region, and that this can lead to spatially heterogenous patterns of reef use by elasmobranchs. This may suggest that including the protection of reefs with different biological characteristics within local conservation strategies may promote rare and vulnerable regional elasmobranch species ranging from stingrays, guitarfishes, reef sharks, and pelagic rays.
-
Efficient deep learning models for diabetic retinopathy screening and severity assessmentDiabetic retinopathy (DR) is a leading cause of preventable blindness worldwide. Early detection through systematic screening is critical, yet manual grading of fundus images by ophthalmologists is labor-intensive, subjective, and increasingly unsustainable given the rising global diabetes burden. This paper presents a dual deep learning approach to automated DR detection, addressing both binary classification (DR vs. No DR) and five-stage severity grading. We propose two complementary models: an ultra-lightweight convolutional neural network with only 11,981 parameters achieving 92% accuracy for binary classification, and a fine-tuned DenseNet121 architecture achieving 87% accuracy for multi-class severity grading. The lightweight model employs depthwise separable convolutions to minimize computational cost while maintaining high diagnostic accuracy, making it suitable for deployment in resource-constrained environments and mobile screening platforms. The DenseNet121 model leverages transfer learning with Mish activation, L2 regularization, and aggressive data augmentation to handle class imbalance on the APTOS 2019 dataset. Our results demonstrate that efficient lightweight architectures can compete with complex models for screening-level tasks, while deeper transfer learning models are essential for detailed clinical severity assessment, providing practical solutions.
-
Invisible identities: Privacy-aware object detection for autonomous vehicles and public surveillanceAnonymising visual data dynamically presents challenges for achieving privacy and security by design, as currently addressed by Privacy-Enhancing Technologies (PETs) and Security-Enhancing Technologies (SETs). The need to decrypt data before applying anonymization or masking techniques introduces potential vulnerabilities, where data could be exposed during processing. This article provides a comprehensive solution to implement robust security measures throughout the entire data lifecycle, including training machine learning models on data without applying differential privacy. Our approach ensures end-to-end encryption to protect data at rest, during processing, and in transit. We encrypt the region of interest in visual imagery at the source, allowing machine learning or deep learning models to use it directly for training purposes. This ensures models focus on abstract visual information rather than private details such as facial features or clothing colour. Our work demonstrates that machine learning models trained on privacy-aware data can be as accurate as those trained on raw data. We employed widely used AES encryption and object detectors like YOLO, Single Shot MultiBox Detector (SSD-MobileNetv1-FPN-640x640, SSD-ResNet101-FPN-640x640), and the EfficientDet family (EfficientDet-D0 and EfficientDet-D1) for detection and classification of visual objects.
-
Novel Jet Stability Evaluating Method for DC Plasma TorchThe jet stability of a DC plasma torch affects not only the service life of the torch but also processing consistency in industrial applications. To evaluate both instantaneous and longstanding jet stabilities of a plasma torch, a novel jet stability evaluation method has been developed in this study. The collected raw signals were first analyzed using the fast Fourier transform and filtered with identified characteristic frequencies. Based on the filtered signals, a 200 ms sliding window method was employed to evaluate the relative fluctuation of arc voltage in terms of both longstanding and instantaneous jet stabilities of the plasma torch. The results show that: (1) the proposed method can effectively evaluate both instantaneous and longstanding jet stability of a DC plasma torch; (2) the arc voltage and arc current signals contain a characteristic frequency, which is strongly influenced by the gas flow rate; (3) the laminar plasma torch operates stably at an arc current of 90 A, and its longstanding jet stability improves with increasing gas flow rate. The findings and proposed method provide informative guidance to those interested in the improvement of plasma jet stability and processing consistency.</jats:p>
-
Developing sensor technology and algorithm for enhancing accuracy of monitoring anomaly sleep at homeSleep apnoea represents one of the most prevalent respiratory disorders worldwide, affecting approximately 49.7% of men and 23.4% of women. While polysomnography (PSG) remains the gold standard for diagnosis, its complexity, high cost, and limited accessibility have motivated the development of alternative diagnostic approaches. As respiration and blood oxygen saturation (SpO2) serve as key indicators in sleep apnoea assessment, photoplethysmography (PPG) has emerged as a promising technology for non-invasive monitoring. This dissertation presents a comprehensive investigation of PPG-based sleep monitoring systems, encompassing hardware development, signal processing algorithms, and advanced machine learning techniques for abnormal sleep detection. The research establishes a systematic framework for wearable sleep monitoring through three key phases of development and validation. First, a non-invasive continuous monitoring system was developed to assess sleep apnoea via SpO2 and heart rate (HR) measurements. Various breathing experiments were performed to investigate the relationship between breathing patterns, SpO2 fluctuations, and HR variations during apnoea events. Motion artifact (MA) analysis revealed directional and frequency dependencies, various adaptive filters were implemented to compare their effectiveness in MA removal. Next, an integrated framework combining variational mode decomposition (VMD) denoising with deep learning autoencoders was developed for automated abnormal SpO2 detection. The study investigated both the relationship between architectural complexity and performance gains, as well as the trade-offs between model complexity and computational requirements. The performance of deep learning autoencoders was evaluated across datasets with different characteristics. The VMD algorithm achieved an SpO2 root-mean-square error (RMSE) of 0.862% against clinical references. Among four autoencoder architectures, the long short-term memory (LSTM) autoencoder performed best (accuracy: 0.86) by capturing temporal SpO2 dynamics during apnoea events, though with higher computational costs. Finally, the feasibility of direct raw PPG signal analysis was investigated, bypassing error-prone SpO2 calculations. A wireless multi-sensor system integrating PPG and accelerometer data with Bluetooth connectivity was developed and evaluated. Comparative analysis of nine machine learning and deep learning models revealed that neural network architectures consistently outperformed traditional methods, with the novel 1D-Branch CNN achieving optimal balance between accuracy (0.92) and computational efficiency (~25% faster than CNN-LSTM). This system demonstrates the feasibility of using raw PPG signals for reliable abnormal sleep detection in clinical applications, emerges as particularly promising, offering an effective compromise between the computational efficiency, high performance and the need for generalization in variable environments.
-
Acceptability and suitability of some Poaceae plants for fall armyworm feeding and ovipositionBackground: The fall armyworm (FAW), Spodoptera frugiperda (Lepidoptera: Noctuidae), has invaded new geographical regions, now spanning Africa, Asia, Oceania and Europe, from its native distribution in the last decade. Little is known about FAW host plants in recently invaded habitats; although more than 300 hosts have been reported in the Americas, its native habitat. In our study, we evaluated the acceptability and suitability of 12 cultivated varieties of plants from Africa, in the family Poaceae, for FAW herbivory and oviposition. Methods: Experiments investigating larval development, no-choice oviposition, and no-choice larval arrestment-feeding were conducted to evaluate the insect´s ability to utilize these plants for survival. Results: We found that Pennisetum ex. Sengerema, Brachiaria brizantha , Brachiaria ex. Mwanza, Panicum maximum ex. Machakos, Melinis minutiflora and S . bicolor cv. Ochuti were unsuitable plants for FAW larvae. In contrast, Zea mays HB WH505, Panicum glaucum Nutrifeed, S . bicolor cv. Serena and P . purpureum were suitable plants and S . bicolor cv. Ochuti was well accepted for egg-laying. However, M . minutiflora was not accepted for egg-laying but retained early instar larvae. S . bicolor cv. Andiwo was not well accepted for egg-laying and S . bicolor cv. Gadam yielded lighter pupae. Conclusions: Our findings provide insights into the performance of FAW larvae on different Poaceae plants and how well they are accepted by FAW female moths for oviposition. We recommend to study further selected plants ( M . minutiflora , S . bicolor cv. Andiwo and S . bicolor cv. Ochuti), as potential trap or repellent plants for different FAW life stages, in choice tests. This knowledge will help to design ecologically based management strategies for FAW in its new habitats in Kenya and beyond.
-
Third-order time stepping methods for superdiffusion using weighted and shifted Grünwald–Letnikov formulae with nonsmooth dataIn this paper, we study a numerical method for the Caputo time fractional wave equation with nonsmooth data. We first introduce a class of third-order approximations, known as weighted and shifted Grünwald-Letnikov approximations, to approximate the Caputo fractional derivative. Based on this, we develop a new time stepping method for solving the time fractional wave equation. After applying corrections to several initial steps, the proposed time stepping method achieves a convergence order of O(k3)$$O(k^3)$$ for nonsmooth data, where k denotes the time step size. We also analyze the stability regions of the proposed time stepping method, which show that the scheme is unconditionally stable for α∈(1,1.94)$$ \alpha \in (1, 1.94) $$, and conditionally stable for α∈[1.94,2)$$ \alpha \in [1.94, 2) $$. Numerical experiments are presented to validate the theoretical findings.
-
Time discretization schemes for stochastic subdiffusion and fractional wave equations with integrated additive noiseIn this paper, we introduce a time discretization scheme for solving the stochastic subdiffusion equation based on the two-fold integral-differential and two step backward differentiation formula (ID2-BDF2). We prove that this scheme attains a convergence rate of O ( τ α + γ − 1 / 2 ) for 1 / 2 < α + γ < 2 with α ∈ (0, 1) and γ ∈ [0, 1]. Our approach regularizes the additive noise through a two-fold integral-differential (ID2) calculus and discretizes the equation using BDF2 convolution quadrature, achieving superlinear convergence in solving the stochastic subdiffusion. Furthermore, we extend the scheme to solve the stochastic fractional wave equation, proving that the scheme achieves a convergence rate of O ( τ min { 2 , α + γ − 1 / 2 } ) for α ∈ (1, 2) and γ ∈ [0, 1]. Numerical examples are presented to validate the theoretical results for the linear problem. The numerical observations further indicate that the same convergence rates also apply to stochastic semilinear time-fractional equations.
-
An efficient reduced-order approximation for the stochastic Allen-Cahn equationIn this paper, we propose and analyze an efficient numerical method for solving the stochastic Allen-Cahn equation with additive noise. The method combines a stabilized semi-implicit time discretization scheme with a reduced-order finite element spatial discretization method. The core idea is to approximate the original high-dimensional solution space via a low-dimensional subspace, constructed by the Proper Orthogonal Decomposition (POD) method based on an ensemble of snapshots from the full-order model at selected time instances. First, we rigorously establish the spatio-temporal strong convergence rates between the mild solution and the reduced-order solution. Second, in large-sample simulations, the reduced-order basis exhibits a certain generalization capability in capturing the average behavior of the numerical solutions. Numerical experiments are provided to verify the theoretical error estimates and to demonstrate the effectiveness of the proposed method.
-
A corrected Crank–Nicolson scheme for the time fractional parabolic integro-differential equation with nonsmooth dataThis paper proposes a corrected Crank–Nicolson (CN) scheme for solving time fractional parabolic integro-differential equations which involve Caputo time fractional derivative and fractional Riemann–Liouville (R-L) integral. The weighted and shifted Grünwald–Letnikov (WSGL) formulae is adopted to approximate the time fractional Riemann–Liouville integral. The Crank–Nicolson scheme is applied to approximate the Caputo time fractional derivative. After appropriating corrections, the proposed scheme attains the optimal convergence order of O(\tau^2) with respect to the time step size \tau for both smooth and nonsmooth data at any fixed time $t$. When combined with the Galerkin finite element method for spatial discretization, it forms a fully discrete scheme. The second-order error estimate for this scheme is rigorously established using the Laplace transform technique and verified by some numerical examples.
-
Dynamic pricing-driven load optimization in islanded microgrid for home energy management systemsIn order to maximize residential energy use and reduce electricity costs, home energy management systems (HEMS), are crucial. A distinct dynamic pricing-driven load optimization technique for HEMS running in islanded mode where grid access is either limited or nonexistent is presented in this paper. Incorporating important distributed energy resources like photovoltaic (PV) systems, electric vehicles (EVs), battery energy storage systems (BESS), and limited grid access, the optimal schedule is also used for real-time dynamic pricing analysis. A planned short-duration outage and a full 24-hour islanding scenario are the two different outage scenarios that are assessed. In short-duration outages, grid dependency decreases to roughly 10% in the spring and approximately 50% in the summer and winter, according to a thorough seasonal analysis conducted in the spring, summer, and winter. The suggested approach guarantees total self-sufficiency in the 24-hour outage scenario, with local resources satisfying load demands in full.
-
Banal deception and human-AI ecosystems: A study of people’s perceptions of LLM-generated deceptive behaviourLarge language models (LLMs) can provide users with false, inaccurate, or misleading information, and we consider the output of this type of information as what Natale calls ‘banal’ deceptive behaviour [53]. Here, we investigate peoples’ perceptions of ChatGPT-generated deceptive behaviour and how this affects people’s behaviour and trust. To do this, we use a mixed-methods approach comprising of (i) an online survey with 220 participants and (ii) semi-structured interviews with 12 participants. Our results show that (i) the most common types of deceptive information encountered were over-simplifications and outdated information; (ii) humans’ perceptions of trust and chat-worthiness of ChatGPT are impacted by ‘banal’ deceptive behaviour; (iii) the perceived responsibility for deception is influenced by education level and the perceived frequency of deceptive information; and (iv) users become more cautious after encountering deceptive information, but they come to trust the technology more when they identify advantages of using it. Our findings contribute to understanding human-AI interaction dynamics in the context of Deceptive AI Ecosystems and highlight the importance of user-centric approaches to mitigating the potential harms of deceptive AI technologies.
-
Numerical algorithms for nonlinear fractional stochastic Volterra-type equationIn this work, we investigate a class of nonlinear stochastic Volterra-type evolution equations, which can be regarded as an extension of the results reported in Qiao et al. (Fract Calc Appl Anal 27:1136–1161, 2024). For such equations, we propose an Euler scheme and rigorously establish the existence, uniqueness, and regularity of the solution. Moreover, we present the detailed numerical implementation of the scheme and derive the corresponding error estimates.
-
Carbon fiber-reinforced piezoelectric nanocomposites: Design, fabrication and evaluation for damage detection and energy harvestingCarbon fiber-reinforced polymers (CFRPs) can be used in aging infrastructures as a reinforcement because of their excellent mechanical properties, and they can also be used in the support maintenance and repair work of these structures. However, the development of CFRPs as reinforcement while achieving self-powered damage detection is still challenging. Herein, a sodium potassium niobate (KNN) nanoparticle-filled epoxy (KNN–EP) plate was fabricated and combined with advanced CFRP electrodes. The obtained composites exhibited dramatically enhanced mechanical properties. In addition, CFRP contributed to the energy harvesting output (peak-to-peak output voltage V p p = 7.25 mV), which was over 600 % higher than that of the KNN–EP plate. Thus, this composite could work as a force sensor for damage detection. In the end-notch bending test, the voltage signals generated by CFRP/KNN–EP composite accurately corresponded to the crack growth, which could provide the real-time crack state and prediction of fracture occurrence. Therefore, this work provided a new strategy for structural enhancement and kinetic energy harvesting, which can be used to detect damage behavior in infrastructures.
-
Numerical approximation for a stochastic time-fractional cable equationAn efficient numerical method is proposed to address a stochastic time-fractional cable equation driven by fractionally integrated additive noise. Under the reasonable assumptions, we rigorously establish for the first time, the existence, uniqueness, and regularity of the mild solution for this equation. For spatial discretization, a semi-discrete scheme is constructed employing the Galerkin FEM, and the optimal spatial error estimate is derived based on the semigroup approach. In temporal discretization, a piecewise constant function is introduced to approximate the noise, leading to the formulation of a regularized stochastic time-fractional cable equation. A detailed proof of the temporal error estimates is provided via the semigroup approach. Numerical experiments demonstrate that the temporal convergence order attains O ( τ 1 / 2 ) for initial data of either smooth or non-smooth type. The order is independent of the parameters α 1 ∈ ( 0 , 1 ) , α 2 ∈ ( 0 , 1 ) , and β ∈ ( 0 , 1 ) in the equation. These results perfectly align with the theoretical predictions.
-
Correction to: Visualization for epidemiological modelling: challenges, solutions, reflections and recommendations (2022) by Dykes et al.In the original version of this article, references 113–120, 123–140 and 143 were incorrectly numbered. This has been corrected on the publisher’s website.
-
Erratum to “Combined heat and power from the intermediate pyrolysis of biomass materials: Performance, economics and environmental impact” [Appl. Energy 191 (2017) 639–652]The publisher regrets that Fig. 3 in Page 643 contains errors in data labels.
-
Correction: Existence, uniqueness and regularity for a semilinear stochastic subdiffusion with integrated multiplicative noiseThe original online version of this article was revised: The co-author’s name was misspelled. The co-author's name was spelled Ziqiang Li but should have been Zhiqiang Li. The original version is corrected.
