Uncertainty Quantification and Causal Considerations for Off-Policy Decision Making
Muhammad Faaiz Taufiq · 2025
Read Uncertainty Quantification and Causal Considerations for Off-Policy Decision Making by Muhammad Faaiz Taufiq (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2502-06011.txt.
Description
Muhammad Faaiz Taufiq presents a thesis on judging new policies from old data in a 2025 work. Off-policy evaluation asks how a fresh rule for choices would have done, using records made under a distinct rule. The method matters where testing a new rule live is costly or risky. Past methods suffer from noisy estimates and from hidden causes that the data never recorded.
Three projects make up the thesis. The first cuts noise by focusing on the spread of outcomes rather than the shift in rules. The second gives prediction ranges with firm finite-sample coverage, for risk-aware choices. The third sets bounds that stay valid even when hidden causes distort the records. Those bounds are used to test digital twin models, flagging where a twin's forecasts drift from real behavior.
Catalog record
| CREATOR | Muhammad Faaiz Taufiq |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 395 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2502-06011