01
Federated & Split Learning
Distributed learning under non-IID data, client heterogeneity, communication limits, and adversarial conditions.
Trustworthy AI · Edge Intelligence
Minh K. Quan
AI researcher · PhD candidate at Deakin University
I build learning systems that can operate across real devices while remaining private, robust, and efficient.
My research spans federated and split learning, ML privacy and security, edge intelligence, and on-device representation learning. My PhD thesis, Trustworthy Federated Learning for Sixth Generation Cyber Physical Systems, studies how distributed AI can remain dependable under heterogeneity, privacy constraints, and adversarial behavior.
I’m especially interested in research that connects theory to systems: ideas that can be implemented, profiled, attacked, defended, and ultimately deployed.
privacy
edge AI
federated
PhD thesis under examination
SystemsEdge & on-device ML
TrustPrivacy, robustness & security
LearningFederated & split learning
research focus
02
Trustworthy & Private AI
Privacy leakage, differential privacy, Byzantine robustness, fairness, causal attribution, and auditable learning systems.
03
Edge & On-device ML
Adaptive model partitioning, streaming audio learning, resource-aware inference, and hardware-conscious ML.
news
Aug 2026
I accepted an offer to join Analog Devices (ADI) in Limerick, Ireland, as a Senior Engineer, AI/ML Software.
Aug 2026
My PhD thesis entered formal examination at Deakin University.
Jun 2026
StreamSplit appeared at ACM MobiSys 2026, exploring continuous audio representation learning across heterogeneous edge devices.
May 2026
Recognized as an ICML 2026 Gold Reviewer.
2025
Domain-Adapted Granger Causality was accepted as a poster at the NeurIPS 2025 Workshop on CauScien.
2025
Our survey on federated learning for cyber-physical systems appeared in IEEE Communications Surveys & Tutorials.
academic service
I contribute to the research community through peer review and technical programme service, with 40+ completed review assignments spanning machine learning, federated learning, privacy, IoT, and communications.
40+
Peer reviews
Reviewer for venues including ICML, IEEE Internet of Things Journal, and IEEE Open Journal of the Communications Society.
10
Springer Nature reviews
10 completed review reports across 6 journals in the last two years (as of Aug 2026), including Scientific Reports, Artificial Intelligence Review, and Cluster Computing.
Gold
ICML 2026 Reviewer
Received the ICML 2026 Gold Reviewer Award in recognition of reviewing contributions.
TPC
IEEE ICC
Technical Programme Committee member for IEEE ICC 2025 and IEEE ICC 2026; also an IEEE OJCS Exemplary Reviewer Award recipient.
Springer Nature reviewing includes Scientific Reports, Artificial Intelligence Review, Cluster Computing, Journal of King Saud University Computer and Information Sciences, Journal of Data, Information and Management, and International Journal of Networked and Distributed Computing. Two additional assignments were active as of August 2026.
selected work
all papers →StreamSplit
6G
under reviewsecurity + causality
Certified Causal Attribution for Real-Time Attack Forensics in 6G Network Slicing
IEEE Transactions on Information Forensics and Security · manuscript under review
FL × CPS
Research style
From theoretical failure modes to deployable systems.
Across my work, I tend to start with a concrete failure mode — privacy leakage, adversarial behavior, resource volatility, or causal ambiguity — then ask what can be proved, measured, and built about it.