Overview
Bianca Zadrozny is a computer scientist (PhD UC San Diego; IBM Research, later industry), known for work on calibrated probability estimation, cost-sensitive learning, and applied machine learning.
Key Contributions / Features
- Calibration by isotonic regression / binning (Zadrozny-Elkan 2002, and Zadrozny-Elkan 2001): non-parametric methods to turn classifier scores into accurate binary and multiclass probabilities.
- Cost-sensitive learning and the "reduction to binary" methodology.
Related