#voronoi

Question 1: Single choice

hardAI-checkedNearest Neighbors & Kernel Methods

To speed up a two-class 1-nearest-neighbor classifier you remove "redundant" training points, whose removal changes the prediction for no possible query. Which condition makes a point redundant?

Its Voronoi cell borders only cells of its own class1
It is farther than average from the training points of the other class2
Its own nearest neighbor has the same class3
It lies inside the convex hull of its own class4

Timed test