Accompanying code for the paper Maximum Likelihood With Bias-Corrected Calibration is Hard-To-Beat at Label Shift Adaptation
Accepted to ICML 2020
Authors: Amr Alexandari*, Anshul Kundaje† and Avanti Shrikumar*†
*co-first authors, †co-corresponding authors
See https://colab.research.google.com/github/kundajelab/labelshiftexperiments/blob/master/notebooks/demo/blog_colab.ipynb for a demo notebook illustrating the core functionality
Core calibration and label shift adaptation code lives in https://github.com/kundajelab/abstention
See the notebooks/ folder for code to replicate tables in the paper, and post a github issue if you have questions!