Risk-Sensitive Reinforcement Learning, Mathematical Finance, and Machine Learning
My research lies at the intersection of risk measurement, decision-making under uncertainty, and machine learning. I focus on developing principled, mathematically grounded models for risk-sensitive reinforcement learning, optimal policy learning, portfolio optimization, and credit risk assessment. These efforts aim to advance both theoretical understanding and the development of practical tools in finance, operations, and intelligent systems.
In a recent project, we proposed a novel Distributional Reinforcement Learning (DRL) algorithm that optimizes a broad class of static spectral risk measures. The method offers theoretical convergence guarantees and clear policy interpretation by leveraging coherent risk decomposition. Empirical results show improved performance over existing risk-neutral and risk-sensitive models on benchmark environments.
Previous work includes ensemble learning models for credit rating prediction using real-world financial data, and sequence-based clustering of firms for long-term credit risk assessment. These projects integrate AI techniques with domain-specific modeling to tackle high-stakes decision problems in finance.
