Question

Which of the following methods can not achieve zero training error on any linearly separable dataset?

a) decision tree

b) 15-nearest neighbors

c) hard-margin svm

d) perceptron

Answer 1

b) 15-nearest neighbors