Theorem 26.13: for separable-with-margin data with ‖x‖ ≤ R, the hard-SVM output has P[y⟨w_S,x⟩ ≤ 0] ≤ 2R‖w⋆‖/√m + (1 + R‖w⋆‖)√(2ln(2/δ)/m) w.p. ≥ 1−δ

Initializing...

Theorem 26.13: for separable-with-margin data with ‖x‖ ≤ R, the hard-SVM output has P[y⟨w_S,x⟩ ≤ 0] ≤ 2R‖w⋆‖/√m + (1 + R‖w⋆‖)√(2ln(2/δ)/m) w.p. ≥ 1−δ · Prove2Me