Journal Articles

Permanent URI for this collectionhttps://mro.massey.ac.nz/handle/10179/7915

Browse

Author: search.filters.author.Susnjak T

Search Results

Now showing 1 - 10 of 37
  • Thumbnail Image
    Item
    Data-driven progress prediction in construction: a multi-project portfolio management approach
    (Frontiers Media S A, 2025-12-16) Tagharobi M; Babaeian Jelodar M; Susnjak T; Mahdiyar A
    Introduction: Construction projects often experience delays and cost overruns, particularly in regions like New Zealand, where natural hazards and climate change exacerbate these risks. Despite extensive research on forecasting overall construction timelines, limited attention has been given to stage-wise progress across the project lifecycle, constraining project managers’ ability to monitor performance and respond to risks. Methods: To address this gap, the study develops a stage-based forecasting model using Multinomial Logistic Regression, which was identified as the most suitable method after comparison with selected machine learning approaches within the study’s scope and assumptions. A stepwise comparative framework was employed to assess combinations of duration, value, type, and contractor involvement, measuring accuracy, log-loss, and Cohen’s kappa using 10 years of New Zealand construction data. Model reliability was further examined using confusion matrices to derive sensitivity, specificity, predictive values, and balanced accuracy. Validation was conducted through cross-validation, ROC/AUC, and temporal hold-out testing. Results: The results show that while all models performed reasonably well, the model using only project duration and value achieved the highest accuracy. The validation procedures confirmed the framework’s robustness and generalisability. Visualisations further illustrated milestone-specific progress predictions (5%–100%), making stage-wise forecasts easy to interpret. Discussion: The model provides project managers with practical insights for planning, monitoring, risk management, and resource allocation. By offering a transparent and interpretable approach, it bridges statistical forecasting with real-world practice, supporting timely delivery and data-driven infrastructure development. Future research could incorporate additional factors, extend the model locally and internationally, and explore integration with digital twins or real-time adaptive systems.
  • Thumbnail Image
    Item
    Generalisation Bounds of Zero-Shot Economic Forecasting using Time Series Foundation Models
    (MDPI (Basel, Switzerland), 2025-12-01) Jetwiriyanon J; Susnjak T; Ranathunga S
    This study investigates the transfer learning capabilities of Time-Series Foundation Models (TSFMs) under the zero-shot setup, to forecast macroeconomic indicators. New TSFMs are continually emerging, offering significant potential to provide ready-trained and accurate forecasting models that generalise across a wide spectrum of domains. However, the transferability of their learning to many domains, especially economics, is not well understood. To that end, we study TSFM’s performance profile for economic forecasting, bypassing the need for training bespoke econometric models using extensive training datasets. Our experiments were conducted on a univariate case study dataset, in which we rigorously back-tested three state-of-the-art TSFMs (Chronos, TimeGPT, and Moirai) under data-scarce conditions and structural breaks. Our results demonstrate that appropriately engineered TSFMs can internalise rich economic dynamics, accommodate regime shifts, and deliver well-behaved uncertainty estimates out of the box, while matching and exceeding state-of-the-art multivariate models currently used in this domain. Our findings suggest that, without any fine-tuning and additional multivariate inputs, TSFMs can match or outperform classical models under both stable and volatile economic conditions. However, like all models, they are vulnerable to performance degradation during periods of rapid shocks, though they recover the forecasting accuracy faster than classical models. The findings offer guidance to practitioners on when zero-shot deployments are viable for macroeconomic monitoring and strategic planning.
  • Thumbnail Image
    Item
    GatedFusion-Net: Per-pixel modality weighting in a five-cue transformer for RGB-D-I-T-UV fusion
    (Elsevier B V, 2026-05-01) Brenner M; Reyes NH; Susnjak T; Barczak ALC
    We introduce GatedFusion-Net (GF-Net), built on the SegFormer Transformer backbone, as the first architecture to unify RGB, depth ( D ), infrared intensity ( I ), thermal ( T ), and ultraviolet ( UV ) imagery for dense semantic segmentation on the MM5 dataset. GF-Net departs from the CMX baseline via: (1) stage-wise RGB-intensity-depth enhancement that injects geometrically aligned D, I cues at each encoder stage, together with surface normals ( N ), improving illumination invariance without adding parameters; (2) per-pixel sigmoid gating, where independent Sigmoid Gate blocks learn spatial confidence masks for T and UV and add their contributions to the RGB+DIN base, trimming computational cost while preserving accuracy; and (3) modality-wise normalisation using per-stream statistics computed on MM5 to stabilise training and balance cross-cue influence. An ablation study reveals that the five-modality configuration (RGB+DIN+T+UV) achieves a peak mean IoU of 88.3 %, with the UV channel contributing a 1.7-percentage-point gain under optimal lighting (RGB3). Under challenging illumination, it maintains comparable performance, indicating complementary but situational value. Modality-ablation experiments reveal strong sensitivity: removing RGB, T, DIN , or UV yields relative mean IoU reductions of 83.4 %, 63.3 %, 56.5 %, and 30.1 %, respectively. Sigmoid-Gate fusion behaves primarily as static, lighting-dependent weighting rather than adapting to sensor loss. Throughput on an RTX 3090 with a MiT-B0 backbone is real-time: 640 × 480 at 74 fps for RGB+DIN+T, 55 fps for RGB+DIN+T+UV, and 41 fps with five gated streams. These results establish the first RGB-D-I-T-UV segmentation baselines on MM5 and show that per-pixel sigmoid gating is a lightweight, effective alternative to heavier attention-based fusion.
  • Thumbnail Image
    Item
    The Erosion of Cybersecurity Zero-Trust Principles Through Generative AI: A Survey on the Challenges and Future Directions
    (MDPI (Basel, Switzerland), 2025-12-01) Xu D; Gondal I; Yi X; Susnjak T; Watters P; McIntosh TR
    Generative artificial intelligence (AI) and persistent empirical gaps are reshaping the cyber threat landscape faster than Zero-Trust Architecture (ZTA) research can respond. We reviewed 10 recent ZTA surveys and 136 primary studies (2022–2024) and found that 98% provided only partial or no real-world validation, leaving several core controls largely untested. Our critique, therefore, proceeds on two axes: first, mainstream ZTA research is empirically under-powered and operationally unproven; second, generative-AI attacks exploit these very weaknesses, accelerating policy bypass and detection failure. To expose this compounding risk, we contribute the Cyber Fraud Kill Chain (CFKC), a seven-stage attacker model (target identification, preparation, engagement, deception, execution, monetization, and cover-up) that maps specific generative techniques to NIST SP 800-207 components they erode. The CFKC highlights how synthetic identities, context manipulation and adversarial telemetry drive up false-negative rates, extend dwell time, and sidestep audit trails, thereby undermining the Zero-Trust principles of verify explicitly and assume breach. Existing guidance offers no systematic countermeasures for AI-scaled attacks, and that compliance regimes struggle to audit content that AI can mutate on demand. Finally, we outline research directions for adaptive, evidence-driven ZTA, and we argue that incremental extensions of current ZTA that are insufficient; only a generative-AI-aware redesign will sustain defensive parity in the coming threat cycle.
  • Thumbnail Image
    Item
    Modeling the Chaotic Semantic States of Generative Artificial Intelligence (AI): A Quantum Mechanics Analogy Approach
    (Association for Computing Machinery, 2025-12-01) Liu T; McIntosh TR; Susnjak T; Watters P; Halgamuge MN
    Generative artificial intelligence (AI) models have revolutionized intelligent systems by enabling machines to produce human-like content across diverse domains. However, their outputs often exhibit unpredictability due to complex and opaque internal semantic states, posing challenges for reliability in real-world applications. In this paper, we introduce the AI Uncertainty Principle, a novel theoretical framework inspired by quantum mechanics, to model and quantify the inherent unpredictability in generative AI outputs. By drawing parallels with the uncertainty principle and superposition, we formalize the trade-off between the precision of internal semantic states and output variability. Through comprehensive experiments involving state-of-the-art models and a variety of prompt designs, we analyze how factors such as specificity, complexity, tone, and style influence model behavior. Our results demonstrate that carefully engineered prompts can significantly enhance output predictability and consistency, while excessive complexity or irrelevant information can increase uncertainty. We also show that ensemble techniques, such as Sigma-weighted aggregation across models and prompt variations, effectively improve reliability. Our findings have profound implications for the development of intelligent systems, emphasizing the critical role of prompt engineering and theoretical modeling in creating AI technologies that perceive, reason, and act predictably in the real world.
  • Thumbnail Image
    Item
    MM5: Multimodal image capture and dataset generation for RGB, depth, thermal, UV, and NIR
    (Elsevier B V, 2026-02-01) Brenner M; Reyes NH; Susnjak T; Barczak ALC
    Existing multimodal datasets often lack sufficient modality diversity, raw data preservation, and flexible annotation strategies, seldom addressing modality-specific cues across multiple spectral channels. Current annotations typically concentrate on pre-aligned images, neglecting unaligned data and overlooking crucial cross-modal alignment challenges. These constraints significantly impede advanced multimodal fusion research, especially when exploring modality-specific features or adaptable fusion methodologies. To address these limitations, we introduce MM5, a comprehensive dataset integrating RGB, depth, thermal (T), ultraviolet (UV), and near-infrared (NIR) modalities. Our capturing system utilises off-the-shelf components, incorporating stereo RGB-D imaging to provide additional depth and intensity (I) information, enhancing spatial perception and facilitating robust cross-modal learning. MM5 preserves depth and thermal measurements in raw, 16-bit formats, enabling researchers to explore advanced preprocessing and enhancement techniques. Additionally, we propose a novel label re-projection algorithm that generates ground-truth annotations directly for distorted thermal and UV modalities, supporting complex fusion strategies beyond strictly aligned data. Dataset scenes encompass varied lighting conditions (e.g. shadows, dim lighting, overexposure) and diverse objects, including real fruits, plastic replicas, and partially rotten produce, creating challenging scenarios for robust multimodal analysis. We evaluate the effects of multi-bit representations, adaptive gain control (AGC), and depth preprocessing on a transformer-based segmentation network. Our preprocessing improved mean IoU from 70.66% to 76.33% for depth data and from 72.67% to 79.08% for thermal encoding, using our novel preprocessing techniques, validating MM5’s efficacy in supporting comprehensive multimodal fusion research.
  • Thumbnail Image
    Item
    The Lyme Disease Controversy: An AI-Driven Discourse Analysis of a Quarter Century of Academic Debate and Divides
    (2025-04-04) Susnjak T; Palffy C; Zimina T; Altynbekova N; Garg K; Gilbert L
  • Thumbnail Image
    Item
  • Thumbnail Image
    Item
    Towards clinical prediction with transparency: An explainable AI approach to survival modelling in residential aged care.
    (Elsevier B.V., 2025-02-18) Susnjak T; Griffin E
    BACKGROUND AND OBJECTIVE: Scalable, flexible and highly interpretable tools for predicting mortality in residential aged care facilities for the purpose of informing and optimizing palliative care decisions, do not exist. This study is the first and most comprehensive work applying machine learning to address this need while seeking to offer a transformative approach to integrating AI into palliative care decision-making. The objective is to predict survival in elderly individuals six months post-admission to residential aged care facilities with patient-level interpretability for transparency and support for clinical decision-making for palliative care options. METHODS: Data from 11,944 residents across 40 facilities, with a novel combination of 18 features was used to develop predictive models, comparing standard approaches like Cox Proportional Hazards, Ridge and Lasso Regression with machine learning algorithms, Gradient Boosting (GB) and Random Survival Forest. Model calibration was performed together with ROC and a suite of evaluation metrics to analyze results. Explainable AI (XAI) tools were used to demonstrate both the cohort-level and patient-level model interpretability to enable transparency in the clinical usage of the models. TRIPOD reporting guidelines were followed, with model parameters and code provided publicly. RESULTS: GB was the top performer with a Dynamic AUROC of 0.746 and a Concordance Index of 0.716 for six-month survival prediction. Explainable AI tools provided insights into key features such as comorbidities, cognitive impairment, and nutritional status, revealing their impact on survival outcomes and interactions that inform clinical decision-making. The calibrated model showed near-optimal performance with adjustable clinically relevant thresholds. The integration of XAI tools proved effective in enhancing the transparency and trustworthiness of predictions, offering actionable insights that support informed and ethically responsible end-of-life (EoL) care decisions in aged care settings. CONCLUSION: This study successfully applied machine learning to create viable survival models for aged care residents, demonstrating their usability for clinical settings via a suite of interpretable tools. The findings support the introduction into clinical trials of machine learning with explainable AI tools in geriatric medicine for mortality prediction to enhance the quality of EoL care and informed discussions regarding palliative care.
  • Thumbnail Image
    Item
    Multimodal Deep Learning for Android Malware Classification
    (MDPI (Basel, Switzerland), 2025-02-28) Arrowsmith J; Susnjak T; Jang-Jaccard J; Buccafurri F
    This study investigates the integration of diverse data modalities within deep learning ensembles for Android malware classification. Android applications can be represented as binary images and function call graphs, each offering complementary perspectives on the executable. We synthesise these modalities by combining predictions from convolutional and graph neural networks with a multilayer perceptron. Empirical results demonstrate that multimodal models outperform their unimodal counterparts while remaining highly efficient. For instance, integrating a plain CNN with 83.1% accuracy and a GCN with 80.6% accuracy boosts overall accuracy to 88.3%. DenseNet-GIN achieves 90.6% accuracy, with no further improvement obtained by expanding this ensemble to four models. Based on our findings, we advocate for the flexible development of modalities to capture distinct aspects of applications and for the design of algorithms that effectively integrate this information.