Theorem 10.2: if every round of AdaBoost has εₜ ≤ 1/2 − γ, the training error of its output is at most exp(−2γ²T)

Initializing...

Theorem 10.2: if every round of AdaBoost has εₜ ≤ 1/2 − γ, the training error of its output is at most exp(−2γ²T) · Prove2Me