We provide two jupyter notebooks each replicates one of the figures in our paper:
- joint_v_staged_model_complex.ipynb replicates Figure 2
- joint_v_staged_sample_complexity.ipynb replicates Figure 3
- python run_halfspaces.py replicates Figure 4
For reference implementation of learning to defer methods, please see our updated code in this repo instead https://github.com/clinicalml/human_ai_deferral