IEEE i-COSTE 2026 - NPS Australia Submission System

THE CONNECTION BETWEEN CLIMATE CHANGE AND HEALTH STATUS IN GLOBAL: ADDRESSING THE SDG 3 AND 13

The study investigates the complex relationship between climate conditions and health status by categorizing countries into distinct clusters, focusing on how climate change disproportionately affects low-income nations despite historical emissions from high-income countries.The study investigates the complex relationship between climate conditions and health status by categorizing countries into distinct clusters, focusing on how climate change disproportionately affects low-income nations despite historical emissions from high-income countries

Another Test Submission by Rashed

Testing

Smell-Augmented Symbolic Regression for Explainable Cross-Project Defect Prediction

Cross-Project Defect Prediction (CPDP) remains a critical challenge in software engineering, primarily due to severe dataset shift and systemic class imbalance between source and target repositories. Furthermore, the state-of-the-art black-box machine learning models have an inherent lack of interpretability and out-of-distribution generalization. This paper presents an empirical framework leveraging Smell-Enhanced Symbolic Regression (SR) to derive explicit, human-readable mathematical formulations for defect prediction, capable of robust extrapolation. The proposed framework is rigorously evaluated across eight heterogeneous software project transitions against 23 baseline configurations using five standardized performance metrics and non-parametric significance analysis. The empirical results demonstrate that Symbolic Regression delivers complete interpretability without sacrificing the predictive performance of complex, top-tier models.

SpecSynLite: Localization-Guided Residual Harmonic Synthesis for Energy-Efficient Real-Time Speech Enhancement

The main contributions are: (1) a hybrid enhancement-synthesis architecture with selective activation; (2) a localization-guided synthesis mask that restricts synthesis to speech-dominant regions; and (3) an energy-aware training objective that explicitly optimizes the quality-per-compute tradeoff for sustainable edge deployment.

Explainable Hybrid Deep Learning Framework for Alzheimer’s Disease Classification using Cross-Attention Fusion

• Novel Attention-Guided Hybrid Framework:In this
paper, we propose highly capable and attention-based
deep learning framework, that combines ConvNeXtV2-
Tiny and EfficientNetV2-S using Cross-Attention Fusion
and Convolutional Block Attention Modules (CBAM) to
achieve accurate classification of the four Alzheimer’s
disease classes.
• High Classification Performance: The proposed frame-
work leads to an excellent overall test accuracy of 97.33%
and a Macro ROC-AUC of 0.9996 indicating highly
robust and reliable performance for automated early de-
mentia detection.
• Explainable AI for Model Interpretation: To improve
model interpretability, we integrate a comprehensive XAI
framework incorporating ten CAM variants, including
Grad-CAM, together with SHAP and LIME to provide
detailed visual explanations of the model’s predictions.
• Deep Latent Space and Misclassification Analysis:
We perform a thorough analysis of boundary cases us-
ing a purpose-built misclassification analysis and t-SNE
and UMAP clustering. These visualizations convert raw
model outputs into clinically meaningful insights about
structural decision boundaries.

Development of a Thematic Approach for Sustainable Green Campuses for Higher Education Institutions (HEIs): Bangladesh Perspective

Higher Education Institutions (HEIs) all around the world have progressively incorporated sustainability into their social, economic, and environmental operations since the Sustainable Development Goals (SDGs) were familiarized in 2015. The use of ecological and smart green construction practices in HEIs has accelerated due to the growing global emphasis on sustainable development. Nonetheless, in Bangladesh, comprehensive mapping of campus-based sustainability initiatives, such as smart green campuses, is still limited. This current study looks into the role of smart green campuses in increasing sustainability in Bangladeshi HEIs by utilizing energy-efficient technology, environmentally responsible materials, and intelligent resource-management systems. Furthermore, the study assesses how well campus sustainability initiatives match with key SDGs, specifically SDG 6 (clean water), SDG 7 (clean energy), SDG 11 (sustainable cities), SDG 12 (responsible consumption), and SDG 15 (life on land). This study utilized an exploratory research to gather insights from high-indexed journals (Scopus, Web of Science, etc.) related to sustainability in HEIs. A thematic model was developed by analyzing 50 papers to identify sustainability assessment indicators. The expected results will exhibit how smart green campuses can improve institutional sustainability performance and act as drivers for more extensive social change.

Testing Paper Submission

Another test

Testing Paper Submission

EfficientNetV2S with End Ensemble for Robust Bangla Handwritten Character Recognition

Recognizing handwritten characters continues to be an important challenge within the domain of image processing. Specifically, Bangla handwritten characters represent a difficult challenge due to their complex shapes, variation, and high interclass similarity. We introduce the EfficientNetV2S model to overcome the difficulties. It is adapted specifically to recognize the complex and visually similar characters in the Bangla language. Research on Bangla handwritten characters is still limited, although it is the world’s 7th most spoken language. This study presents a deep learning approach that was trained and tested on a dataset containing handwritten Bangla character samples. The model can properly extract the features of the images and classify the images correctly with high accuracy. This approach enabled the model to learn complex features of the images and achieve remarkable character recognition accuracy. We have increased the number of categories of the different characters by combining the “BanglaLekha-Isolated” and “Matrivasa-raw (Ekush)” datasets, which are used for training our model. Our
method achieves an impressive 96.63% accuracy, which proves our proposed technique works very effectively and can be trusted. The results confirm that the end-ensemble technique solves recognition challenges accurately. Our technique can offer strong potential for real-world applications such as automation and education. This work significantly advances the field of Bangla character recognition and encourages further exploration.

Conference Site: IEEE i-COSTE 2026