Local & Global explanation Interpretable vs. Powerful 用一个modelA解释(模拟结果)另一个modelB Linear模拟DNN——LIME(local explanation) decision-tree