Machine Learning in Financial Markets
Professor: Tomoyuki Ichiba
Description: This undergraduate research project proposes evaluating the application of Tabular Foundation Model (TabPFN) in the context of finance (e.g., robust portfolio optimization, credit scoring and financial anomaly detection). Financial tabular datasets frequently present extreme class imbalance and market distribution shifts that degrade the predictive accuracy of traditional models over time. By leveraging TabPFN’s in-context learning capabilities, class-imbalance-aware dataset distillation will be integrated to compress large financial tables while preserving rare default signals. This study aims to deliver a privacy-conscious, highly scalable machine learning framework and the corresponding mathematical problems for real-time risk assessment across financial portfolios.
Preferred Qualifications:
