Dr. Seyedmostafa Safavi Trustworthy AI · Cybersecurity

Research & publications

Advancing trustworthy AI and cyber defense.

My research grounds human-centric, trustworthy AI in rigorous statistical and data-science methodology, from privacy-preserving federated learning and explainability to the robustness of AI in healthcare and digital finance.

Google Scholar profile Research site →

Research themes

Index / 01–06
/ 01

Explainability & Fairness

Making AI decisions transparent, fair and accountable, with calibrated uncertainty and human-AI collaboration.

/ 02

Federated & Private Learning

Differential privacy, secure aggregation and distributed estimation for sensitive, multi-organization data.

/ 03

Healthcare & Critical Systems AI

Privacy-preserving diagnostics and models that remain robust under real-world distribution shift.

/ 04

AI Governance & Regulatory Science

Evaluation protocols, assurance cases, auditing and policy frameworks for responsible AI.

/ 05

AI-Driven Cybersecurity

Anomaly detection, behavioral analytics and trust-building frameworks for security decision-making.

/ 06

Adversarial Robustness

Attack taxonomies and defence mechanisms for AI in medical imaging and other high-stakes settings.

Grants & projects

Selected research funding

~MYR 800K total

HORIZON 2020 · MYR 300,000 · 2020–2022

Explainable Federated Learning (XFL) for Multi-Hospital Fetal Health Assessment

Principal Investigator · European Commission

FRGS · MYR 150,000 · 2014–2017

Privacy Framework for Health Information on Wearable & Portable Devices

Principal Investigator · National Research Grant

INTERNATIONAL GRANT · MYR 180,000 · 2023–2025

Explainable AI (XAI) and Cybersecurity: Trust-Building Frameworks

Co-Investigator

CREST GRANT · MYR 200,000 · 2021–2024

Securing the Sustainable Future of Manufacturing

Co-Investigator

Selected publications

Peer-reviewed work

A complete, categorised publication list is available on request and via Google Scholar.

Safavi, S., & Shukur, Z. (2014). Conceptual privacy framework for health information on wearable devices. PLoS ONE, 9(12), e114306.

Q1

Safavi, S., Abdulnabi, M. S. H., Rana, M. E., & Alizadeh, S. (2025). From black box to trustworthy AI: A secure framework for explainable cybersecurity decision-making. 2025 Int. Conf. ASSIC, 1–4. IEEE.

Mohan, M. H., Seeboruth, K., Rana, M. E., Umar, U. S., Mohan, T., & Safavi, S. (2025). Enhancing fetal health assessment: Automated head circumference measurement via deep learning segmentation. 2025 ASSIC, 1–8. IEEE.

Chandran, A. L., Samual, J., Safavi, S., & Ali, A. (2025). A comparative analysis of machine learning models for detecting malware in Android devices. Journal of Cyber Security and Risk Auditing, 4, 327–346.

EL Bakkali, J., EL Bardouni, T., Safavi, S., et al. (2016). Behaviors of percentage depth dose curves: A Monte Carlo Geant4 study. Radiation Physics and Chemistry, 125, 199–204.

Q2

Safavi, S., Shukur, Z., & Razali, R. (2013). Reviews on cybercrime affecting portable devices. Procedia Technology, 11, 650–657. Elsevier.

Safavi, S., & Shukur, Z. (2015). CenterYou: A permission-based privacy framework (pseudo-data technique) in Android. Malaysian Patent No. 710420-12-5,412.

Patent

Forthcoming / under review

In the pipeline

Adversarial Robustness in Medical-Imaging AI

Attack taxonomies and defence mechanisms. Q1, in process.

Dual-Phase Brain-Tumour Segmentation

Gumbel-Softmax + cascaded Swin Transformer framework. Q1, accepted.

Human-AI Collaboration for Anomaly Detection

Behavioral analytics in cybersecurity. Q2, in process.

Research collaboration

Interested in collaborating?

I welcome research partnerships, co-supervision and joint projects across trustworthy AI, federated learning and cybersecurity.

Reach out