Policy learning from action-inclusive feedback on OpenML K ≥ 3
57.98Policy AccuracyCB Policy
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| CB Policy2022.06 | 57.98 | — | — | |
| AI-IGLFeedback=Action-Inclusive2022.06 | 35.74 | — | 0.59 | |
| IGL (full CI)Feedback=full CI2022.06 | 15.65 | — | 0.3 | |
| Constant Action2022.06 | — | 25.28 | — |