Sustainable Engineering and Innovation

Holistic generative and discriminative models for intrusion detection: A GAN-assisted multiclass classification mechanism

Abdullah Albalawi — Wed, 22 Apr 2026 00:00:00 +0000

Traditional intrusion detection systems often struggle with the complexity of modern, multi-dimensional cyber threats. This study proposes a hybrid four-phase methodology that integrates unsupervised Generative Adversarial Network (GAN)-based anomaly scoring with supervised multiclass classification for attack type and severity. Utilizing a dataset of 40,000 network records, the framework employs domain-specific feature engineering, including payload analysis and z-score normalization. A GAN trained on 11,934 normal samples generated discriminator-based anomaly scores to serve as probabilistic inputs for subsequent models. While the GAN alone showed limited binary detection performance (AUC-ROC=0.4983), it provided valuable features for the hybrid architecture. In the multiclass classification phase, BiLSTM achieved the highest overall accuracy (34.3%), while Random Forest demonstrated superior binary performance (AUC-ROC=1.0000). The results highlight the inherent challenges of threat categorization in imbalanced, real-world datasets. The study concludes that while GANs are ineffective as standalone classifiers, their discriminator outputs function effectively as probabilistic features within a unified framework. This approach bridges a gap in IDS research by combining generative modeling with dual-task classification for more robust network security.

Predictive analytics of battery use conditions and degradation: A data-driven approach

Khowshik Dey, Serkan Varol — Wed, 15 Apr 2026 00:00:00 +0000

Lithium-ion batteries are the critical components of the latest energy storage devices, ranging from consumer electronics to electrically powered automobiles. Despite their significant applications in the storage of electrical energy devices, the functionality of these batteries largely relies on the nature of the charge/discharge cycles. Batteries subjected to complete charge/discharge cycles deteriorate faster than batteries that undergo partial cycles, which limits their possibility for reuse and recycling. Understanding and predicting battery consumption conditions can substantially help with lifecycle management and sustainable recycling initiatives. This study investigates a data-driven approach for predicting the use condition of lithium-ion batteries based on degradation criteria. Logistic regression, Naïve Bayes, and decision tree models are used to predict the previously used condition of a battery based on several variables such as days of degradation, energy throughput, C/10 capacity, and state of health (SoH). The dataset is collected from the existing literature and preprocessed to extract the required variables. After preprocessing and variable selection, the models are made and tested with cross-validation and ROC analysis. The analysis of the results suggests that the decision tree classifier performs better compared to other classification models in terms of accuracy, F1 score, and AUC values. The findings prove the applicability of the predictive analysis technique in assisting battery lifecycle management by categorizing the batteries that have been used in different scenarios for recharging.

CNN technique for brain tumor detection and classification using MRI

Wed, 15 Apr 2026 00:00:00 +0000

In general, an uncontrolled and sudden growth of cells poses a significant threat to human life, particularly in the brain region. Diagnosis of these tumors by classifying and dividing them to determine the location, structure, and proportion of tumors is a major challenge, despite the strenuous efforts made by researchers in this field. In this study, statistical image processing techniques and computational intelligence were used to suggest several approaches for recognizing brain tumors and cancer. In this research, the CNN algorithm was used for classifying brain cancer images into two categories: cancerous and non-cancerous. The image features are extracted by entering data through the first layer and gradually moving to the other layers until reaching the final layer. In this work, the CNN algorithm, ReLu, and Maxpool are used with three steps of filters (16,32,64), the Adam technique is used for stochastic optimization, and the SoftMax function for classification is implemented. Kaggle dataset for 7023 patient images is used, the network is trained until reach 0verall accuracy 63.74% at epoch 35, with a learning rate of 0.003.

Optimizing network performance through AI-driven media-to-text conversion

Ali Hussein Alnooh, Nawar A. Sultan, Ali Yasir Kuti — Thu, 09 Apr 2026 00:00:00 +0000

One of the most challenging issues in networks and cloud computing is the overhead caused by the exchanged data. Several approaches in the literature have been proposed to mitigate this issue. However, there is still a lack of innovative methods and techniques to reduce the network's overhead. Hence, this work proposes an AI-driven method that converts media data (e.g., images, audio, or videos) into text to enhance the overall network performance. The proposed method is evaluated based on bandwidth savings, throughput, and latency. The findings demonstrate that the proposed method achieves a 98% bandwidth reduction and a 3.6 times higher throughput, with high accuracy (BLEU-4 > 0.78 for captions, WER < 12% for speech). Moreover, the statistical validation shows a significant improvement in latency (150ms for audio and 950ms for video) and a packet loss rate of 0.3%. Finally, the proposed method is considered adaptable to IoT, edge computing, and cloud systems due to its cost-effectiveness.

Enhancing user interaction through the design of digital information environments in information services

Tue, 07 Apr 2026 00:00:00 +0000

The interface computer systems are adaptive and context-aware in order to enhance user engagement in online information environments, such as Ukrainian digital services, such as Diia, and library systems based on VuFind. The discovery of the usability perceptions, critical design aspects, and cultural influences is conducted through a qualitative thematic analysis through an analytical sample and in-depth interviews with 10-20 users of digital platforms in Ukraine. The study proposes a context-aware UX framework (CAUXF), which has adaptive interface design, environmental, behavioural, and cognitive-emotional characteristics, and user modelling with personalisation. The proposed model consists of a user context profile, a context information manager, a conversation supervisor, a context-awareness service manager, and embedded mobile services. The researchers concluded that 45% of digital platform users in Ukraine rated them as being moderately challenging to use, half of them recognised key elements of design such as simplified user interfaces, and 60% desired the capability to customise it. The cultural and contextual analysis of the Ukrainian-language interfaces in the current study revealed that 80% of them were preferred. The paper highlights the relevance of user-centred design considerations that meet the needs of Ukrainian digital users in their technological and cultural contextual environment by using more simplified interfaces, mobile compatibility, language localisation, and infrastructure issues.

Adhesively bonded joints and their applications

Fehim Findik — Tue, 31 Mar 2026 00:00:00 +0000

Adhesively bonded joints offer a superior alternative to traditional mechanical fasteners, providing uniform stress distribution, reduced weight, and the ability to join dissimilar materials. Their versatility makes them indispensable across the aerospace, automotive, and medical industries. However, performance is highly dependent on adhesive selection, surface preparation, and environmental resilience. This study provides a comprehensive critique of failure modes, including interfacial, cohesive, and mixed-mode damage, often triggered by temperature fluctuations or cyclic loading. It further examines standardized testing protocols under ASTM, ISO, SAE, and EN frameworks, such as lap shear and fatigue testing, to ensure structural reliability. A significant portion of the research focuses on sustainability, addressing the challenges of toxicity and recyclability through bio-based materials and reversible bonding. Finally, the study explores emerging innovations like self-healing polymers and AI-assisted selection, which are set to revolutionize the future of adhesive technology. By balancing structural efficiency with environmental considerations, adhesively bonded joints remain a cornerstone of modern engineering.

Sustainable urban heritage strategies for Iraq's holy cities: A case study of the Old City of Najaf

Haider Majid Hasan, Husam Sachit Senah, Adil Mahdi Jabbar — Thu, 19 Mar 2026 00:00:00 +0000

Najaf is a preeminent historical sacred city, hosting millions of secular visitors annually. However, it suffers from chronic heritage management inefficiencies due to the lack of context-sensitive integrated systems. This study innovatively synthesizes the Historic Urban Landscape (HUL) approach with Sustainable Development Goals (SDGs) to redefine urban heritage in Old Najaf as a liveable, adjustable landscape rather than mere historical remnants. By focusing on the historic center, the research presents a framework for landscape-based planning tailored to sacred sites with deep religious and political resonance. Specifically, the study aligns urban interventions with SDG 11 (Sustainable Cities and Communities) to establish a measurable route for integrated heritage plans. Drawing on expert consensus, three sustainable development models were identified, integrating UN Principles, SDG 11 indicators, and UNESCO’s HUL aspects. These models provide a strategic balance between modern social infrastructure needs and the preservation of sacred historical identity, offering a replicable blueprint for similar global sacred contexts.

Morlet wavelet–based olfactory-evoked EEG features for random forest classification of normal, aMCI, and Alzheimer’s disease

Nabila A. Alsharif — Tue, 17 Feb 2026 00:00:00 +0000

Olfactory impairment and abnormal frontal EEG oscillations are recognized as early markers of Alzheimer’s disease (AD). Using a publicly available olfactory EEG dataset of 35 subjects spanning normal cognition, amnestic mild cognitive impairment (aMCI), and AD, each with MMSE scores and demographics, stimulus-locked epochs from four electrodes (Fp1, Fz, Cz, Pz) were processed with wavelet-based time–frequency analysis. Band-limited power ratios (delta, theta, alpha, beta) were computed as log-transformed post-odor/baseline values and aggregated to subject-level features. Statistical analyses revealed graded attenuation of odor-evoked frontal (Fp1) band-power ratios across groups, with significant differences in several band–odor combinations. PCA of Fp1 features showed partial separation of diagnostic categories, while multi-channel features offered weaker discrimination. Random forest classifiers trained on Fp1-only features achieved 66.7% test accuracy, outperforming the four-channel model (55.6%), with moderate sensitivity, specificity, and precision. These findings highlight that compact frontal wavelet-derived band-power ratios during olfactory stimulation carry diagnostically relevant information for distinguishing Normal, aMCI, and AD. The transparent pipeline, combining time–frequency processing, subject-level aggregation, and multiclass classification, offers a scalable framework that can be extended to larger cohorts or integrated with multimodal biomarkers.

Hybrid AI model-driven dynamic spectrum sharing for 6G wireless IoT networks

Tue, 17 Feb 2026 00:00:00 +0000

The immense scale of the Internet of Things growth in 6G is utterly inconceivable to address utilizing conventional static spectrum allocations. A paradigm shift towards dynamic spectrum sharing is necessitated. In this article, a hybrid artificial intelligence model that combines deep reinforcement learning and a blockchain-based distributed consensus engine has been presented. Intelligent, secure, and efficient spectrum sharing may be accomplished using our model. The proposed methodology employs multi-agent reinforcement learning for efficient decentralized decision-making and IoT-enabled spectrum utilization. Specifically, IoT devices can use MARL to dynamically determine their power budget or spectrum resources to avoid inducing or experiencing interference while delivering acceptable quality of service. Using a blockchain engine to record and validate spectrum transactions enables transparent security in spectrum access. Our proposed hybrid AI model may be used to improve spectrum efficiency by 35%-40% while lowering energy usage by around 30% via intelligent sleep-wake lexicography methodologies and decision predication relative to traditional 5G. We thoroughly covered the spectrum management topic in 6G-IoT, demonstrating the feasibility of AI-based solutions.

The application of blockchain technologies in information security and computer systems data

Yurii Shevchuk, Yevhenii Tytarchuk, Serhii Zybin, Anton Sorokun, Taras Khometa — Mon, 16 Feb 2026 00:00:00 +0000

The rapid expansion of digital systems and the increasing frequency of cyberattacks have made information and data security a critical global concern. This challenge is particularly severe in Ukraine, where prolonged conflict with Russia has involved hybrid warfare, including persistent cyberattacks on digital and information infrastructures. This study examines the use of blockchain technology to improve secure data management through an intelligent Hybrid Blockchain–Relational (HBR) architecture. Sensitive data are stored on a private blockchain (Hyperledger Fabric), while less sensitive data are maintained in a relational database (PostgreSQL), with data integrity ensured through Merkle root anchoring. A simulation using Ukraine’s Land Cadaster data served as the case study. Under Byzantine fault and system degradation conditions, Blockchain-based Consensus Optimization (BRCO) achieved a 40% reduction in transaction completion time and a 66.7% increase in node fault tolerance compared to Practical Byzantine Fault Tolerance (PBFT). The proposed HBR+BRCO design demonstrated low latency (50 ms), efficient resource usage, and a throughput of 500 TPS, highlighting its effectiveness and real-world applicability.

Detecting spatial and temporal myopia using machine learning algorithms

Mon, 16 Feb 2026 00:00:00 +0000

This study aims to examine the ability of machine learning algorithms to detect strategic myopia in organizations. As it consists of two variables, the first machine learning algorithms as independent variable with two dimensions: Decision trees classification and K- Means clustering, while the second variable is strategic myopia as dependent variable with two dimensions: spatial and temporal myopia. This study adopted a quantitative approach, and a publicly available HR dataset obtained from Kaggle was used to ensure data privacy. The dataset, which has been used in this study, represents the organizational internal factors with 14,999 employees’ records. Both decision trees and K-means were applied to the internal factors’ datasets, showing the likelihood of employees staying in the organizations and clustering the customers into three clusters. The study revealed that both decision trees and k-means can help organizations in detecting spatial and temporal myopia, and the researchers recommended that organizations should integrate machine learning algorithms in their decision-making processes.

A computer vision as a tool for automated quality control in smart manufacturing

Fri, 02 Jan 2026 00:00:00 +0000

Computer vision (CV) has emerged as one of the most significant enablers of intelligent factoring system quality control, automated in the context of the AI revolution in the industrial setting today. In this research, we discuss how CV-based architecture can be applied to achieve real-time, adaptive, and scalable quality assurance. This is new research because it is an amalgamation – the evaluation of different mathematical models and artificial intelligence (AI). Deep learning, transfer learning, Bayesian networks, and edge computing are among the solutions, as are fog-cloud partnerships and their direct impact on manufacturing output, productivity, and decision-making efficiency. The article provides comparative data on the performance of other CV frameworks in different industrial conditions by critically examining the new case studies. The practical implications are recommendations for adopting vision-driven systems to improve product consistency, increase human-machine interaction, and reduce operational downtime. In addition, the paper identifies shortcomings in computational resources, system compatibility, and information security that should be addressed in the next generation of smart factories.

Estimating survival rates using artificial intelligence combined with the Aalen–Johansen estimator in multi-state models

Hasanain Jalil Neamah Alsaedi, Fatema S. Al-Juboori, Ruqaia Jwad Kadhim — Thu, 18 Sep 2025 00:00:00 +0000

Accurate survival prediction is essential for clinical decision-making, health economics, and treatment planning. Traditional methods like the Kaplan-Meier and Cox models are widely used but have limitations when applied to complex multi-state processes or individualized predictions. The Aalen–Johansen estimator, a non-parametric approach suited for multi-state Markov models, improves population-level inference but lacks the ability to incorporate covariates or capture nonlinear relationships. In this study, we propose a hybrid framework that combines the Aalen–Johansen estimator with artificial intelligence (AI) techniques, specifically gradient boosting machines (GBM) and long short-term memory (LSTM) networks. By transforming transition probabilities into subject-level pseudo-observations, AI models can learn personalized survival functions based on individual covariates. We validate our approach on both simulated and real-world clinical datasets. The hybrid model outperforms traditional estimators in predictive accuracy, as measured by calibration and discrimination metrics such as Brier score and area under the curve (AUC). This AI–Aalen–Johansen framework enhances risk stratification and clinical decision-making by providing more accurate, scalable, and interpretable survival predictions. Our results support its potential as a valuable tool in modern healthcare analytics, contributing to the advancement of precision medicine.