Optimal weighted nearest neighbour classifiers

Explore this paper's citation graph

Summary

An asymptotic expansion for the excess risk (regret) of a weighted nearest-neighbour classifier is derived, and it is argued that improvements in the rate of convergence are possible under stronger smoothness assumptions, provided the authors allow negative weights.

Type
article
Published
2011-01-30
Cited by
291
References
41
Access
Open access

Keywords

Nearest neighbour, Classifier (UML), Regret, Pattern recognition (psychology), Asymptotically optimal algorithm

References

Cited by

Related papers

No related papers recorded.