FX
PhD Researcher in Computer Science

Flavjo Xhelollari

Trustworthy AI · Quantum Machine Learning · Privacy & Robustness

I study secure, private, robust, and verifiable machine-learning systems, spanning hybrid quantum–classical models and structured world models, with an emphasis on reliable behavior under real deployment constraints.

2 first-author journal articles 4 peer-reviewed workshop papers
Email GitHub LinkedIn

Research

My work sits at the intersection of trustworthy machine learning and quantum computing, with an emphasis on privacy, robustness, and formal guarantees.

Trustworthy AI
Quantum Machine Learning
Privacy-Preserving ML
Certified & Adversarial Robustness
Formal Verification

Publications

Selected work on privacy-preserving QML, robust hybrid quantum–classical learning, formal verification, and structured world models for physical AI.

Journal articles

Ensembling Personalized Quantum Models with Local Differential Privacy

Published · 2026
Flavjo Xhelollari, Samuel Yen-Chi Chen, Junaid Farooq, Juntao Chen
Quantum Machine Intelligence
Personalized label-local differential privacy for user-specific quantum models with privacy-aware ensemble aggregation.
Quantum ML Local Differential Privacy Ensemble Learning IBM Quantum
DOI
Framework for personalized quantum models with local differential privacy View framework

Robustness of Hybrid Classical-Quantum Transfer Learning Models Under Differential Privacy Constraints

Published · 2026
Flavjo Xhelollari, Ying Mao, Samuel Yen-Chi Chen, Juntao Chen
ACM Transactions on Quantum Computing
A controlled classical-versus-PQC transfer-learning study of differential privacy, adversarial robustness, and hardware-noise sensitivity.
Transfer Learning Differential Privacy Adversarial Robustness Hybrid QML
DOI
Hybrid classical-quantum transfer-learning framework View framework
Workshop papers

Where Structure Enters a World Model: From Predictive Fit to Physical Rollout

Accepted · NeurIPS 2026
Panayiotis Christou, Satyapragny Kar, Flavjo Xhelollari, Parijat Dube, Rajarshi Das
NeurIPS 2026 Workshop on World Models in Physical AI
Studies where structured inductive bias should enter world models, combining temporal, equivariant, and mechanics-aware structure to improve physical rollout and sparse-force adaptation.
World Models Physical AI Physics-Informed ML Equivariance
Workshop
Physical rollout comparison for NTP, force-law, angular-momentum, and TDV structured world models View figure

Relational Certificates for End-to-End Verification of Hybrid Quantum–Classical Machine-Learning Pipelines

Accepted · QCE 2026
Flavjo Xhelollari, Tao Li, Juntao Chen
4th International Workshop on Quantum Machine Learning: From Research to Practice · IEEE Quantum Week / QCE 2026 · Toronto
End-to-end relational certificates for robustness, stability, and deployment reliability of composed hybrid QML pipelines.
Formal Verification Relational Logic Reliability Hybrid QML
End-to-end relational verification framework for hybrid QML View framework

Transferable Certified Robustness for Hybrid Quantum–Classical Machine Learning

Accepted · QCE 2026
Flavjo Xhelollari, Tao Li, Juntao Chen
PRQAI: Towards Privacy and Robustness in Quantum Artificial Intelligence · IEEE Quantum Week / QCE 2026 · Toronto
Transfer-aware robustness certification that explicitly accounts for finite-shot uncertainty and simulator-to-device drift.
Certified Robustness Sim-to-Hardware NISQ Finite Shots
Hybrid QML classifier with offline and online transfer-aware certified robustness workflows View framework

Ensembling Personalized Quantum Models with LDP via Meta-Learning

AAAI 2025
Flavjo Xhelollari, Juntao Chen
QC+AI Workshop at AAAI 2025 · Philadelphia
Quantum ML Meta-Learning LDP Personalization

Recent Updates

Sep 2026 Where Structure Enters a World Model: From Predictive Fit to Physical Rollout accepted at the NeurIPS 2026 Workshop on World Models in Physical AI.
Aug 2026 Ensembling Personalized Quantum Models with Local Differential Privacy accepted for publication in Quantum Machine Intelligence.
Aug 2026 Robustness of Hybrid Classical-Quantum Transfer Learning Models Under Differential Privacy Constraints accepted for publication in ACM Transactions on Quantum Computing.
Jul 2026 Two first-author workshop papers accepted at IEEE Quantum Week / QCE 2026 in Toronto.
Feb 2025 Presented Ensembling Personalized Quantum Models with LDP via Meta-Learning at the QC+AI Workshop at AAAI 2025.

Travel & Adventures

Travel photo
Travel photo
Travel photo
Travel photo
Travel photo
Travel photo