Adaptive Variational Quantum Reinforcement Learning for Dynamic Portfolio Optimization
Dr. Lei Zhang, Department of Information Systems.
Samuya Brahmbhatt, Department of Information Systems.
Extending the baseline QRL portfolio optimization paper by introducing regime-awareness into the quantum circuit architecture. The baseline trains one fixed VQC and deploys it regardless of market conditions. Hidden Markov Model is used to detect market regimes from return data, run an ablation study to identify which circuit configuration performs best per regime, and deploy a switching mechanism that routes to the appropriate pre-trained circuit when the HMM detects a regime shift. The core hypothesis is that different market conditions — calm bull periods versus volatile crisis periods — benefit from different quantum circuit topologies, and that exploiting this leads to more stable portfolio performance during market stress periods. This work also serves as empirical motivation for future research on Quantum Meta-Reinforcement Learning, where instead of switching between pre-trained circuits, a single circuit learns to adapt rapidly to new regimes.
Identifying software vulnerabilities to quantum attacks
Dr. Lei Zhang, Department of Information Systems.
Existing public-key encryption algorithms are vulnerable to quantum attacks, yet they are widely used in software systems that process and store sensitive data. To protect the data again quantum attacks, we are investigating an automated approach to detect the software vulnerabilities (i.e., public-key encryption encryption) in existing software (e.g., VPNs) using word embeddings and machine learning techniques.
Emerging software technologies
Dr. Lei Zhang, Department of Information Systems.
My research focuses on the interactions between software engineering and emerging technologies, e.g., quantum computing, machine learning, and natural language processing. Around quantum computing, I am interested in post-quantum cryptography and quantum software engineering. About machine learning, I am interested in applying artificial neural networks to improve software engineering practices. Last but not least, I study applications of natural language processing throughout software development life cycles.
Automated Quantum Issue Labeling Using LLMs
Dr. Lei Zhang, Department of Information Systems.
This project is for GPU hyperparameter tuning for DistilBERT/RoBERTa and GPT classifiers.