This preview shows page 1. Sign up to view the full content.
Unformatted text preview: ume that is between r=1-ε
and r=1. Why do you expect this to work better? Vd ( r ) = k d · r d The required fraction is: Vd (1)
Related fact: Vd (1
Vd (1) ✏) =1 (1 ✏)d The ratio of the volume of the unit sphere and unit cube tends
to 0 as d goes to inifnity.
23 6 10/21/13 k-NN k-NN Use the closest k neighbors to make a decision instead of a
single nearest neighbor Use the closest k neighbors to make a decision instead of a
single nearest neighbor Choose the label that occurs among the majority of the k
nearest neighbors Choose the label that occurs among the majority of the k
nearest neighbors Can produce confidence scores. How? Other refinements: an example’s vote is inversely proportional
to its distance NN vs k-NN Nearest neighbor regression
How do turn k-NN into a regression method? The data Using
1 NN Using
white regions occur
where there ties
Images from http://en.wikipedia.org/wiki/K-nearest_neighbors_algorithm 28 7 10/21/13 Distances and kernels Distances and kernels The Euclidean dista...
View Full Document
- Fall '08
- Machine Learning