top of page

Research Projects | Methods & Other

I lead a number of applied research projects in collaboration with partner organizations, advise undergraduate and Master's students, and pursue my own research in healthcare and public policy.

Beyond Deep Learning: Do Foundation Models Win on Time Series Classification?

January 2026 - Ongoing

Team Members: Ryan Dong, Zander Rhodes, Dylan Xia, Lucas Zheng, Bruno Paes Leao, Arberie Hakaj, and Olympia Brikis

Time Series Foundation Models (TSFMs) pre-trained on large corpora promise to classify industrial time series without task-specific training, but their practical value relative to conventional classifiers remains unclear. We ask: when can zero-shot TSFMs serve as practical substitutes for task-specific narrow classifiers under limited and imperfect industrial data conditions? To answer this, we evaluate six models—three zero-shot TSFMs (Chronos-2, MOMENT, Mantis) and three baselines (XGBoost with tsfresh features, a 1D CNN, and TabPFN)—across 15 industrial datasets spanning seven application domains.

Project Partners: Siemens AI Lab

Other Past Projects

January 2018 - December 2018

  • Statistical Monitoring of Queueing Network

©2021 by yarenbilgekaya

bottom of page