Cost-sensitive retraining via posterior learning debt Journal: Knowledge-Based Systems
Harrison E. Katz · 2026
Read Cost-sensitive retraining via posterior learning debt Journal: Knowledge-Based Systems by Harrison E. Katz (2026) 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-2604-06438.txt.
Description
Katz frames retraining as a cost-touchy regret decision. Deployed systems use fixed dates, though age varies. The watch state is posterior learning debt, the gap from shadow to frozen beliefs. Refresh cost is set against expected regret of waiting. A continuous version retrains when expected regret exceeds cost. A proof-of-concept tests it. An age-tuned debt-threshold policy beats dates refresh in all 72 cells and CUSUM in 58. Debt-utility and hybrid-utility policies beat refresh. The contribution is a transparent layer, not a drift detection swap.
Catalog record
| CREATOR | Harrison E. Katz |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 23001 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2604.06438