2026
CLM-AI-003608<= L_A <= (1+eta)L_* + ln(N)/etaUPPER BOUNDThe cited theorem establishes L_A <= (1+eta)L_* + ln(N)/eta under the recorded model.Source: A decision-theoretic generalization of on-line learning and an application to boosting — Weighted-majority expert regret bound ↗
PROVEN