Sample Complexity of Calibrated Multiaccuracy Depends on Test-Time Randomization
Aaroth · x · 2026-11-03
A theory thread on calibrated multiaccuracy: multicalibration supports omniprediction, fast agreement, and hard core set construction, but is often overkill—global calibration plus multiaccuracy suffices. The author notes a counterintuitive result: calibrated multiaccuracy does require fewer samples (as long as there aren't too many groups), yet its sample complexity depends on whether the model is allowed to randomize at test time—unlike most single-distribution learning problems.
More from Research
- Skill++ paper: routing-aware skill refinement lifts agent routing from 36.4% to 76.3% — dair_ai · 2026-11-03
- CMU launches first Protostar Challenge on climate risk AI, open until Nov 13 — evijit · 2026-11-03
- Tivadar Danka publishes a complete math roadmap for machine learning: linear algebra, calculus, probability — TivadarDanka · 2026-11-03
- Formalizing CFSG Uncovers 96 Math Errors in 47 Peer-Reviewed Papers, Including Annals of Mathematics — skdh · 2026-11-03
- SEC-Bench Pro: frontier labs test agents on 183 real V8 and SpiderMonkey vulnerabilities — NielsRogge · 2026-11-03
- PhD thesis from Aengus Lynch shows aligned AI stays vulnerable: 89% jailbreak, 96% blackmail — StephenLCasper · 2026-11-03