One-nearest-neighbour classification ($1$-NN)

One-nearest-neighbour classification assigns a query the label of its single nearest training observation. Under local regularity, its limiting conditional error is $1-{\sum }_k{p}_k(x{)}^2$.