School of Engineering, Computer and Mathematical Sciences - Te Kura Mātai Pūhanga, Rorohiko, Pāngarau

Permanent link for this collectionhttps://hdl.handle.net/10292/553

AUT is home to a number of renowned research institutes in engineering, and computer and mathematical sciences. The School of Engineering, Computer and Mathematical Sciences strong industry partnerships and the unique combination of engineering, computer and mathematical sciences within one school stimulates interdisciplinary research beyond traditional boundaries. Current research interests include:

Artificial Intelligence; Astronomy and Space Research;

Biomedical Technologies;

Computer Engineering; Computer Vision; Construction Management;

Data Science;

Health Informatics and eHealth;

Industrial Optimisation, Modelling & Control;

Information Security;

Mathematical Sciences Research; Materials & Manufacturing Technologies;

Networking, Instrumentation and Telecommunications;

Parallel and Distributed Systems; Power and Energy Engineering;

Software Engineering; Signal Processing; STEM Education;

Wireless Engineering;

Browse

Now showing 1 - 20 of 1814

SIFT-SNN for Traffic-Flow Infrastructure Safety: A Real-Time Context-Aware Anomaly Detection Framework
(MDPI AG, 2026-01-31) Rathee, Munish; Bačić, Boris; Doborjeh, Maryam
Automated anomaly detection in transportation infrastructure is essential for enhancing safety and reducing the operational costs associated with manual inspection protocols. This study presents an improved neuromorphic vision system, which extends the prior SIFT-SNN (scale-invariant feature transform–spiking neural network) proof-of-concept by incorporating temporal feature aggregation for context-aware and sequence-stable detection. Analysis of classical stitching-based pipelines exposed sensitivity to motion and lighting variations, motivating the proposed temporally smoothed neuromorphic design. SIFT keypoints are encoded into latency-based spike trains and classified using a leaky integrate-and-fire (LIF) spiking neural network implemented in PyTorch. Evaluated across three hardware configurations—an NVIDIA RTX 4060 GPU, an Intel i7 CPU, and a simulated Jetson Nano—the system achieved 92.3% accuracy and a macro F1 score of 91.0% under five-fold cross-validation. Inference latencies were measured at 9.5 ms, 26.1 ms, and ~48.3 ms per frame, respectively. Memory footprints were under 290 MB, and power consumption was estimated to be between 5 and 65 W. The classifier distinguishes between safe, partially dislodged, and fully dislodged barrier pins, which are critical failure modes for the Auckland Harbour Bridge’s Movable Concrete Barrier (MCB) system. Temporal smoothing further improves recall for ambiguous cases. By achieving a compact model size (2.9 MB), low-latency inference, and minimal power demands, the proposed framework offers a deployable, interpretable, and energy-efficient alternative to conventional CNN-based inspection tools. Future work will focus on exploring the generalisability and transferability of the work presented, additional input sources, and human–computer interaction paradigms for various deployment infrastructures and advancements.
Standalone DC Microgrids: Planning, Operation and Uncertainty Management
(Elsevier BV, 2026-01-20) Jayasinghe, H; Gunawardane, K; Hossain, MA; Zamora, R
Standalone power systems in remote areas have traditionally relied on continuously operating fossil fuel generators, leading to high operational costs, reduced efficiency, and substantial carbon emissions. Standalone direct current (DC) microgrids have emerged as a promising alternative due to their lower conversion losses, improved integration of renewable energy sources (RES), and enhanced compatibility with modern DC-native loads and storage technologies. Despite these advantages, the planning, operation, and uncertainty management of standalone DC microgrids remain technically challenging. Intermittent RES generation, stochastic load behaviour, lack of mature standards, and complex control requirements introduce significant design and operational challenges. While numerous studies have proposed methods to address issues in sizing, optimisation, control, energy management, and uncertainty management, a comprehensive and structured review that connects these aspects across the full lifecycle of DC microgrid development is still lacking. This article addresses this gap by providing a systematic review of the state-of-the-art in planning methodologies, operational strategies, and uncertainty management techniques for standalone DC microgrids. The review synthesises theoretical frameworks and practical implementations, critically evaluates existing approaches by identifying their strengths and limitations, and highlights the interdependencies among planning, real-time operation, and uncertainty mitigation. Finally, the article outlines key research challenges and future opportunities to support the reliable, cost-effective, and sustainable deployment of standalone DC microgrids. The novelty of this study lies in its integrated perspective spanning planning, operational control, and uncertainty management, offering valuable guidance for researchers, system designers, and practitioners.
Optimizing the Resistivity of Colloidal SnO₂ Thin Films by Ion Implantation and Annealing
(Elsevier BV, 2024-10-28) Yusuf, AS; Markwitz, M; Chen, Z; Ramezani, M; Kennedy, JV; Fiedler, H
Tin oxide (SnO₂) is a critical material for a wide range of applications, such as in perovskite solar cells, gas sensors, as well as for photocatalysis. For these applications the transparency to visible light, high availability, cheap fabrication process and high conductivity of SnO₂ benefits its commercial deployment. In this paper, we demonstrate that the resistivity of widely colloidal SnO₂ can be reduced by noble gas ion beam modification. After low energy argon implantation with a fluence of 4×10¹⁵ at.cm⁻² at 25keV and annealing at 200°C in air, the resistivity of as-deposited film was reduced from (178±6)μΩcm to (133±5)μΩcm, a reduction of 25%. Hall effect measurements showed that the primary cause of this is the increase in carrier concentration from (8.1±0.3)×10²⁰ cm⁻³ to (9.9±0.3)×10²⁰ cm⁻³. Annealing at 200°C resulted in the removal of defect clusters introduced by implantation, while annealing at 300°C resulted in the oxidation of the films, increasing their resistivity. The concentration of oxygen vacancy defects can be controlled by a combination of low energy noble gas ion implantation and annealing, providing promising performance increases for potential applications of SnO₂ where a low resistivity is crucial.
Lateralized Learning for Multi-class Visual Classification Tasks
(Institute of Electrical and Electronics Engineers (IEEE), 2026-02-03) Siddique, Abubakar; Browne, Will N; Grimshaw, Gina M
The majority of computer vision algorithms fail to find higher-order (abstract) patterns in an image so they are not robust against adversarial attacks. Deep learning considers each input pixel in a homogeneous manner such that different parts of a locality-sensitive hashing table are often not connected, meaning higher-order patterns are not discovered. Hence, these systems are sensitive to noisy, irrelevant, and redundant data, leading to wrong predictions with high confidence. Adversarial attacks exploit this vulnerability by generating deceptive inputs that mislead AI systems. In contrast, human vision is rarely susceptible to adversarial attacks. Vertebrate brains afford heterogeneous knowledge representation through lateralization, enabling modular learning at different levels of abstraction. This work aims to verify the effectiveness, scalability, and robustness of a lateralized approach to real-world problems that contain noisy, irrelevant, and redundant data. Two well-known and widely used adversarial attacks, the Fast Gradient Sign Method and the Iterative Adversarial Technique, are applied to generate corrupted test images. The experimental results on multi-class (200 classes) image classification tasks demonstrate that the proposed system effectively captures hierarchical knowledge representations, enhancing its robustness. Crucially, the lateralized system outperformed four state-of-the-art deep learning-based systems for the classification of normal and adversarial images by 19.05% − 41.02% and 1.36% − 49.22%, respectively.
System Development and Evaluation for Mass Casualty Incidents Triage with Virtual Reality and Artificial Intelligence
(ISCRAM, 2025-05-02) Xia, Peng; Ruan, Ji; Parry, David; Aiello, Stephen; Yu, Jian; Britnell, Sally
This study investigates the integration of Virtual Reality (VR) and Artificial Intelligence (AI) to enhance pre-hospital triage training for Mass Casualty Incidents (MCIs). Traditional training methods, such as field drills and full-scale simulations, are often costly and logistically challenging, while simpler methods like tabletop exercises remain limited in realism. To address these limitations, a VR learning tool was developed to simulate realistic emergency scenarios, providing emergency healthcare professionals with an immersive and cost-effective training environment to refine triage skills. The VR learning tool records both VR sensor data and speech data, and then utilizes statistical and AI methods (such as automatic speech recognition, and natural language processing) to process these data for evaluation. The survey results showed that participants with varying levels of experience found the VR training highly immersive and engaging. Additionally, AI-driven analysis of speech data from the training demonstrated improved consistency and correctness in participants’ communication over time. This research demonstrates VR’s potential as a valuable supplement to traditional training, identifying key areas for future development
Model-based Workflow for Sustainable Production of High-Quality Spirits in Packed Column Stills
(Elsevier BV, 2024-10-28) Díaz-Quezada, S; Wilson, DI; Pérez-Correa, JR
This study addresses water scarcity in Chilean distilleries by developing a model-based engineering workflow. Prolonged droughts, likely driven by global warming, have intensified this problem. The primary challenge is maintaining high-quality spirits while reducing cooling water and energy use during batch distillation in packed columns. This process was modeled using mass and energy balances, resulting in a system of partial differential algebraic equations (PDAE). Our workflow includes mechanistic modeling, disturbance modeling, multiobjective optimization, model predictive control, and Monte Carlo simulations. Our findings show that cooling water consumption can be reduced by up to 35 % and energy consumption by up to 14.4 % while maintaining product quality. The proposed system is robust against operational disturbances and model mismatch, ensuring consistent distillate quality. This research demonstrates the integration of model-based optimization and control strategies in batch distillation processes, which can be replicated in other fruit wine distillation processes for improved sustainability.
Augmenting Knee Rehabilitation Replays Using MediaPipe Pose Estimation
(GitHub, 2024-12-31) Bacic, Boris; Vasile, Claudiu; Ciucă, Marian G.; Feng, Chengwei
Overview This project implements a methodology for analysing knee angles from video footage using MediaPipe Pose estimation. It provides near real-time visualisation and analysis of knee movements, supporting both front-view and side-view perspectives. Note Near real-time visualisation of MP4 videos depends on the computer specification. Hint: If using a decade old laptop for fast scrolling and A-B sequence analysis, consider capturing first a screen cast of the generated augmented video replay. As a stand-alone application, this source code is the second part of the three stage processing workflow, which is also a part of it's parent project codebase 1 intended for home use and advancements of near-future analytical healthcare systems. See more: "Towards nation-wide analytical healthcare infrastructures: A privacy-preserving augmented knee rehabilitation case study".
An Adaptive Levy Flight Chicken Swarm Optimization with Differential Evolution for Function Optimization Problem
(The Science and Information Organization, 2025) Liu, WJ; Zain, AM; Bin Talib, MS; Ma, SJ
This study proposes an improved swarm algorithm, Adaptive Levy Flight Chicken Swarm Optimization with Differential Evolution (ALCSODE), to overcome the low convergence accuracy and imbalance between exploration and exploitation in the original CSO algorithm. The method incorporates adaptive perturbation based on individual differences and a differential evolution mechanism into the rooster update process. An elitism preservation strategy is also applied to enhance population stability and information sharing. The algorithm is evaluated on 24 benchmark functions, including unimodal, high-dimensional multimodal, and CEC2022 functions. Performance metrics such as search trajectories and convergence curves are used to assess its effectiveness. Experimental results show that ALCSODE achieves a better exploration–exploitation trade-off and shows statistically superior performance over seven classical algorithms, confirming its potential as an effective tool for solving complex optimization problems.
Navigating the Ethical and Societal Impacts of Generative AI in Higher Computing Education
(arXiv, 2025-11-19) Mak, Janice; Nakatumba-Nabende, Joyce; Clear, Tony; Clear, Alison; Albluwi, Ibrahim; Andrei, Oana; Angeli, Lorenzo; MacNeil, Stephen; Oyelere, Solomon Sunday; Rattigan, Matthew Hale; Sheard, Judy; Zhu, Tingting
Generative AI (GenAI) presents societal and ethical challenges related to equity, academic integrity, bias, and data provenance. In this paper, we outline the goals, methodology and deliverables of their collaborative research, considering the ethical and societal impacts of GenAI in higher computing education. A systematic literature review that addresses a wide set of issues and topics covering the rapidly emerging technology of GenAI from the perspective of its ethical and societal impacts is presented. This paper then presents an evaluation of a broad international review of a set of university adoption, guidelines, and policies related to the use of GenAI and the implications for computing education. The Ethical and Societal Impacts-Framework (ESI-Framework), derived from the literature and policy review and evaluation, outlines the ethical and societal impacts of GenAI in computing education. This work synthesizes existing research and considers the implications for computing higher education. Educators, computing professionals and policy makers facing dilemmas related to the integration of GenAI in their respective contexts may use this framework to guide decision-making in the age of GenAI.
LIGO Core-Collapse Supernova Detection Using Convolution Neural Networks
(arXiv, 2024-10-08) Pan, Zhicheng; Zahraoui, El Mehdi; Cabrera-Guerrero, Guillermo; Maturana-Russel, Patricio
Core-Collapse Supernovae (CCSNe) remain a critical focus in the search for gravitational waves (GWs) in modern astronomy. Their detection and subsequent analysis will enhance our understanding of the explosion mechanisms in massive stars. This paper investigates a combination of time-frequency analysis tools with convolutional neural network (CNN) to enhance the detection of GWs originating from CCSNe. The CNN was trained on simulated CCSNe signals and Advanced LIGO (aLIGO) noise in two instances, using spectrograms computed from two time-frequency transformations: the short-time Fourier transform (STFT) and the Q-transform. The algorithm detects CCSNe signals based on their time-frequency spectrograms. Our CNN model achieves a near 100% true positive rate for CCSNe GW events with a signal-to-noise ratio (SNR) greater than 0.5 in our test set. We also found that the STFT outperforms the Q-transform for SNRs below 0.5.
Ideological Isolation in Online Social Networks: A Survey of Computational Definitions, Metrics, and Mitigation Strategies
(arXiv, 2026-01-21) Wang, Xiaodan; Liu, Yanbin; Wu, Shiqing; Zhao, Ziying; Hu, Yuxuan; Li, Weihua; Bai, Quan
The proliferation of online social networks has significantly reshaped the way individuals access and engage with information. While these platforms offer unprecedented connectivity, they may foster environments where users are increasingly exposed to homogeneous content and like-minded interactions. Such dynamics are associated with selective exposure and the emergence of filter bubbles, echo chambers, tunnel vision, and polarization, which together can contribute to ideological isolation and raise concerns about information diversity and public discourse. This survey provides a comprehensive computational review of existing studies that define, analyze, quantify, and mitigate ideological isolation in online social networks. We examine the mechanisms underlying content personalization, user behavior patterns, and network structures that reinforce content-exposure concentration and narrowing dynamics. This paper also systematically reviews methodological approaches for detecting and measuring these isolation-related phenomena, covering network-, content-, and behavior-based metrics. We further organize computational mitigation strategies, including network-topological interventions and recommendation-level controls, and discuss their trade-offs and deployment considerations. By integrating definitions, metrics, and interventions across structural/topological, content-based, interactional, and cognitive isolation, this survey provides a unified computational framework. It serves as a reference for understanding and addressing the key challenges and opportunities in promoting information diversity and reducing ideological fragmentation in the digital age.
Bayesian Regularization for Dynamical System Identification: Additive Noise Models
(MDPI AG, 2025-11-14) Niven, Robert K; Cordier, Laurent; Mohammad-Djafari, Ali; Abel, Markus; Quade, Markus
Consider the dynamical system ẋ = ƒ (x), where x ∈ Rⁿ is the state vector, ẋ is the time or spatial derivative, and ƒ is the system model. We wish to identify unknown ƒ from its time-series or spatial data. For this, we propose a Bayesian framework based on the maximum a posteriori (MAP) point estimate, to give a generalized Tikhonov regularization method with the residual and regularization terms identified, respectively, with the negative logarithms of the likelihood and prior distributions. As well as estimates of the model coefficients, the Bayesian interpretation provides access to the full Bayesian apparatus, including the ranking of models, the quantification of model uncertainties, and the estimation of unknown (nuisance) hyperparameters. For multivariate Gaussian likelihood and prior distributions, the Bayesian formulation gives a Gaussian posterior distribution, in which the numerator contains a Mahalanobis distance or “Gaussian norm”. In this study, two Bayesian algorithms for the estimation of hyperparameters—the joint maximum a posteriori (JMAP) and variational Bayesian approximation (VBA)—are compared to the popular SINDy, LASSO, and ridge regression algorithms for the analysis of several dynamical systems with additive noise. We consider two dynamical systems, the Lorenz convection system and the Shil’nikov cubic system, with four choices of noise model: symmetric Gaussian or Laplace noise and skewed Rayleigh or Erlang noise, with different magnitudes. The posterior Gaussian norm is found to provide a robust metric for quantitative model selection—with quantification of the model uncertainties—across all dynamical systems and noise models examined.
Designing with, Not For: Addressing AI Bias through Community-Led Co-Design in Heart Failure Care
(AIS eLibrary, 2025-12-03) Chung, Claris; Henchard, Sandra; Hong, Yvonne; Li, Yuming
Artificial Intelligence (AI) for healthcare holds immense promise but carries a profound risk of amplifying existing health inequities, particularly for underserved groups like Pacific peoples in New Zealand. Standard AI models can perpetuate and scale social and clinical biases that lead to poorer health outcomes. This paper argues that to build equitable AI, we must move beyond purely technical fixes and adopt a new methodology grounded in community partnership. We propose an Equity-Centred Co-Design Framework that directly targets the sources of bias. Using the development of an AI-powered management system for Pacific heart failure patients as a case study, we demonstrate how this framework is applied. By integrating Pacific worldviews, our approach ensures that the community's lived experience shapes the AI's development from its foundation. This paper offers a practical roadmap for researchers and developers to create AI systems that are trustworthy, culturally responsive, and grounded in social justice principles.
Precision and Approximation in Digitisation and Transformation of the Individual: Balancing Accuracy and Well-Being in AI-Driven Digital Systems
(AIS eLibrary, 2025-12-03) Li, Yuming; Sengupta, Pooja; Bajaj, Ruhi; Chung, Claris; Sundaram, David
Chronic diseases are largely caused by unhealthy lifestyle choices and behaviours. Early diagnosis and transformative management of chronic diseases are vital for the well-being of the global population. Unfortunately, data regarding the lifestyle choices and behaviours of individuals are sparse, fragmented, or nonexistent. These problems motivated the question of whether we can use both rough and precise data on individuals in a complementary fashion to diagnose and manage chronic diseases, ultimately leading to the well-being and transformation of the individual. We develop a holistic Measure, Model, Manage framework and an AI-driven granularity adaptation framework that learns interpretable mappings between rough self-reported lifestyle data and precise clinical indicators. Using both publicly available datasets and AI-generated synthetic datasets, we compare the robustness of models across varying input granularities. We demonstrate that chronic disease risk can be accurately predicted using not only high-precision biometric inputs but also rough, qualitative data.
Personalising Nutrition and Lifestyle Recommendations: Analysis of Gene-test Reports by Individual and Geographic Differences
(SAGE Publications, 2026-01-23) Chua, Serene; Mohaghegh, Mahsa; Paul, Sharad P; Miranda, Victor
Introduction: Advances in nutrigenomics have enabled exploration of how genetic variation may relate to nutrition and lifestyle traits. However, the extent to which demographic factors influence the distribution of such variants remains underexplored. Objective: This study examined gender- and region-specific variation in diet- and lifestyle-related genetic traits and described patterns of trait clustering within a cohort of direct-to-consumer gene-test clients. Methods: A cross-sectional analysis was conducted on 503 anonymised gene-test reports covering 41 nutrition- and lifestyle-linked genetic components. Chi-square tests assessed demographic differences in allele frequency distributions. Hierarchical clustering and principal component analysis were applied as exploratory tools to visualise trait patterns. Results: Most individuals exhibited typical genotype distributions, though some demographic differences were observed. Statistically significant gender variation was noted in omega-3/6 metabolism (p = 0.0378). Lactose intolerance showed the greatest regional disparity, disproportionately affecting Asian (p < 0.00001). Marked regional differences were also observed in vitamin-D status (p = 0.0137), omega-3 metabolism (p = 0.0215), pain tolerance (p = 0.0279), fat utilisation (p = 0.0406) and gluten sensitivity (p = 0.0411). Clustering grouped 41 components into 14 sets. Three principal clusters explained 44-80% of the variance. Predictive modelling was limited by incomplete data and class imbalance. Conclusion: This exploratory study highlights modest demographic differences in allele frequencies and demonstrates clustering of nutrition-related genetic traits within a direct-to-consumer dataset. Findings should be interpreted as descriptive signals rather than prescriptive guidance. Future research incorporating phenotypic, biomarker, and outcome data is needed to evaluate functional and clinical significance.
Immune Antibodies Recognizing the Stem Region of SARS-CoV-2 Spike Protein: Molecular Modeling and in Vitro Study of Synthetic Peptides Presentation to the Antibodies
(Elsevier BV, 2025-10-12) Aliper, ET; Ryzhov, IM; Obukhova, PS; Tuzikov, AB; Galanina, OE; Ziganshina, MM; Sukhikh, GT; Krylov, NA; Henry, SM; Efremov, RG; Bovin, NV
Antibodies to peptide 1147 (amino acids 1147–61) of the SARS-CoV-2 S protein are highly diagnostic. Peptide 1147, although located in a region that is partly spatially hidden in the intact protein, is not subject to mutations, suggesting therapeutic potential. The aim of this study was to elucidate the architecture of this region and the way in which it is presented to antibodies. As a model system, this peptide carrying a single lipophilic tail and the same peptide carrying a lipophilic tail at both ends (pseudocyclic) were incorporated into a lipid membrane. Isolated anti-1147 antibodies interacted with it regardless of how the peptide was presented, be that freely exposed via the N-terminus, organized as a pseudocycle, or adsorbed on the surface. Molecular dynamics simulations showed that peptide 1147 is capable of closely approaching the membrane. Analysis of the surface properties of peptide 1147 in membrane-bound states and in particular functional conformations in the full-sized S protein reveals an interface for interaction with antibodies. Interestingly, the latter bears similarities to one published peptide-antibody complex. However, these antibodies, in spite of their high diagnostic significance, show no virus-neutralizing activity, indicating that peptide 1147 has no therapeutic value as a synthetic vaccine.
Dynamic TRM Estimation with Load–Wind Uncertainty Using Rolling Window Statistical Analysis for Improved ATC
(MDPI AG, 2026-02-05) Edeh, Uchenna Emmanuel; Lie, Tek Tjing; Mahmud, Md Apel
The rapid integration of renewable energy sources (RES), particularly wind, together with fluctuating demand, has introduced significant uncertainty into power system operation, challenging traditional approaches for estimating Transmission Reliability Margin (TRM) and Available Transfer Capability (ATC). This paper proposes a fully adaptive TRM estimation framework that leverages rolling-window statistical analysis of net-load forecast errors to capture real-time uncertainty fluctuations. By continuously updating both the confidence factor and window length based on evolving forecast-error statistics, the method adapts to changing grid conditions. The framework is validated on the IEEE 30-bus system with 80 MW wind (42.3% penetration) and assessed for scalability on the IEEE 118-bus system (40.1% wind penetration). Comparative analysis against static TRM, fixed-confidence rolling-window, and Monte Carlo Simulation (MCS)-based methods shows that the proposed approach achieves 88.0% reliability coverage (vs. 81.8% for static TRM) while providing enhanced transfer capability for 31.5% of the operational day (7.5 h). Relative to MCS, it yields a 20.1% lower mean TRM and a 2.5% higher mean ATC, with an adaptation ratio of 18.8:1. Scalability assessment confirms preserved adaptation (12.4:1) with sub-linear computational scaling (1.82 ms to 3.61 ms for a 3.93× network size increase), enabling 1 min updates interval.
Frequency-aware Spatio-temporal Topology Learning for Skeleton-based Human Activity Recognition
(Elsevier BV, 2026-07-01) Xia, Y; Yongchareon, S; Lutui, R; Sheng, QZ
Skeleton-based human activity recognition (HAR) has made significant progress through graph convolutional networks (GCNs) and Transformer architectures for spatiotemporal modeling. However, existing methods either employ predefined static graph topologies that cannot adapt to heterogeneous skeleton data or learn dynamic topologies based solely on local spatiotemporal features, thereby overlooking the global temporal frequency features of joint movements that are important for discovering semantically meaningful spatial relationships. We propose Frequency-Aware Topology Learning Graph Convolutional Network (FATL-GCN), a novel architecture that integrates frequency-aware temporal context to guide adaptive learning of spatial topology. Our approach leverages Time-to-Vector linear frequency encoding to capture both periodic and non-periodic motion patterns, employs frequency-guided topology learning to generate action-specific graphs through temporal-context-driven attention, and incorporates hierarchical multi-scale fusion for robust feature extraction across scales. Extensive experiments achieved top-1 accuracies of 93.8% (cross-subject) and 97.5% (cross-view) on NTU-60, 91.9% (cross-subject) and 93.1% (cross-setup) on NTU-120, and 51.7% on Kinetics-Skeleton. Ablation studies confirm the critical role of our components, with removing the dynamic graph topology causing a 3.5% accuracy drop and removing frequency-aware encoding causing a 2.1% drop.
Digital Transformation in the Public Sector: A Systematic Mapping Study From an Agile Perspective
(Edward Elgar Publishing, 2026-01-13) Payomrat, Nobpo; Senapathi, Mali; Madanian, Samaneh
This chapter presents the results of a systematic mapping study that analysed existing research to provide an overview of the current research landscape regarding implementing digital transformation (DT) in the public sector from an agile perspective. We reviewed 22 papers published between January 2014 and April 2024. Our findings indicate that DT studies in the public sector primarily focus on developing new organisational capabilities and routines. Most publications originate from Europe and the USA, while Asia has contributed the least. The reported benefits of integrating agile practices into DT include improved efficiency, transparency, and adaptability. The main challenges include stakeholder engagement and collaboration issues, bureaucratic and cultural barriers, difficulties with adaptability and change management, and challenges related to policies and regulations.
Generative AI-Enhanced Robust Semantic Communication Architecture for UAV Image Transmission
(IEEE, 2026-02-02) Liu, Canpu; Zhou, Li; Deng, Xinfeng; Zhang, Yichi; Li, Nan; Xiong, Jun; Seet, Boon-Chong
Unmanned aerial vehicle (UAV) wireless image transmission has gained widespread application across various fields due to its flexibility, yet it faces critical challenges such as resource constraints and degradation of reconstruction quality caused by harsh channel conditions. To address these issues, we designed a lightweight semantic communication backbone network that substantially reduces the computational and storage overhead of UAVs through codebook assistance and efficient encoder-decoder design. On this basis, to tackle severe image degradation under adverse channel conditions, we introduced a generative artificial intelligence-based (GAI) enhancement module. Specifically, we developed a semantic refinement network (SRN) that employs an innovative signal-to-noise ratio (SNR) adaptive feature-wise linear modulation (FiLM) layer to dynamically adjust its refinement strategy based on real-time channel quality, fundamentally transforming the image reconstruction paradigm from traditional signal recovery to conditional content generation. Extensive experimental results demonstrate that our proposed framework significantly outperforms the current state-of-the-art method under extreme channel conditions, highlighting its great potential for achieving robust UAV image transmission in challenging operational environments.

Browse

Recent Submissions