Unformatted text preview: refer to Disp simply as the p norm. What happens when p goes to infinity? Dis 1 (x, y) = max xi
i We get the Chebyshev distance
Peter Flach (University of Bristol) yi  Machine Learning: Making Sense of Data August 25, 2012 225 / 349 Figure 8.3 7 (left) Lines connecting points at orderp Minkowski distance 1 from the origin for (from
8 inside) p = 0.8; p = 1 (Manhattan distance, the rotated square in red); p = 1.5; p = 2 (Euclidean distance, the violet circle); p = 4; p = 8; and p = 1 (Chebyshev distance, the
blue rectangle). Notice that for points on the coordinate axes all distances agree. For the
other points, our reach increases with p ; however, if we require a rotationinvariant
2 distance metric then Euclidean distance is our only choice. (right) The rotated ellipse xT RT S2 Rx = 1/4; the axisparallel ellipse xT S2 x = 1/4; and the circle xT x = 1/4. 10/21/13
8. Distancebased models p.235 Deﬁnition 8.2: Distance metric Properties of a distance The Mahalanobis distance
8. Distancebased models Figure 8...
View
Full Document
 Fall '08
 Anderson,C
 Machine Learning, Distance, Nearest neighbor search, University of Bristol, nearest neighbor, Peter Flach

Click to edit the document details