2026
CLM-AI-003603>= m >= (log|H| + log(1/delta))/epsilonLOWER BOUNDThe cited theorem establishes m >= (log|H| + log(1/delta))/epsilon under the recorded model.Source: A theory of the learnable — Finite-class consistent PAC sample bound ↗
PROVENSufficient realizable-case sample size for a consistent learner over a finite hypothesis class.
The question, scope, and sources behind this Registry record.
Under the recorded learning theory model, establish m >= (log|H| + log(1/delta))/epsilon.
Current frontiers derived from accepted Claims.
Assertions tied to evidence, attribution, and review.
The frontier as it changed over time.
Only accepted Claims matching the current specification contribute to the displayed bounds. Strict inequalities remain open; contradictory Claims require editorial review.
No accepted machine-checked reproductions are recorded for this Limit.
Limits Registry. LR-003603. Finite-class consistent PAC sample bound. 2026.