The paper establishes an analog of well-known Novikoff’s theorem on the perceptron learning algorithm’s finite convergence in linearly separated classes. We obtain a similar result concerning the nearest neighbor ...