Publications

Illustration representing life satisfaction prediction and explainable AI

Predicting life satisfaction using machine learning and explainable AI

Alif Elham Khan, Mohammad Junayed Hasan, Humayra Anjum, Sifat Momen. Heliyon, Cell Press, 2024.

This study uses a government survey of approximately 19,000 respondents to develop an interpretable machine-learning framework for predicting life satisfaction. The model achieved 93.80% accuracy and a 73.00% macro F1-score, while identifying a compact set of 27 explanatory questions.

PaperCode
Illustration of an open book, multimodal student data, and transparent learning analytics

Explainable Multimodal Learning Analytics: Transformer-Era Methods for Transparent Student Success and Well-Being

In Explainable Artificial Intelligence and Interpretable Machine Learning: Bridging Data Science and Educational Inquiry. CRC Press, 2026, pp. 52–84. In press.

ISBN: 978-1-041-21740-4.

Chapter proof
Illustration of Shadow Loss projection geometry

Shadow loss: Memory-linear deep metric learning for efficient training

Alif Elham Khan, Mohammad Junayed Hasan, Humayra Anjum, Nabeel Mohammed. arXiv preprint.

Shadow Loss performs metric-learning comparisons in a compact projection space, reducing memory and computation while preserving class structure. The method is model-agnostic and was evaluated across balanced, imbalanced, medical, and general image datasets.

arXiv
Illustration of hierarchy-preserving medical representation learning

Climbing the label tree: Hierarchy-preserving contrastive learning for medical imaging

Alif Elham Khan. arXiv preprint, 2025.

This work introduces Hierarchy-Weighted Contrastive and Level-Aware Margin objectives that use medical label taxonomies directly during representation learning. Across medical-imaging benchmarks, the framework improves representation quality while better preserving relationships in the label hierarchy.

arXiv
Illustration of a multi-qubit quantum circuit

Quantum Energy Teleportation across Multi-Qubit Systems using W-State Entanglement

Alif Elham Khan, Humayra Anjum, Mahdy Rahman Chowdhury. arXiv preprint, 2025.

This work develops a multi-qubit quantum-energy-teleportation protocol using W-state multipartite entanglement. Three-, four-, and five-qubit circuits were evaluated on noiseless simulators and IBM superconducting hardware.

arXivCoverage

Research and applied AI

Interested in collaborating?

I’m open to research collaborations and selected technical partnerships in agentic, multimodal, explainable, and resource-efficient AI, particularly for healthcare and education.

Email me Google Scholar