Please use this identifier to cite or link to this item:
https://hdl.handle.net/2440/63428
Citations | ||
Scopus | Web of ScienceĀ® | Altmetric |
---|---|---|
?
|
?
|
Type: | Conference paper |
Title: | BoostML: An adaptive metric learning for nearest neighbor classification |
Author: | Zaidi, N. Squire, D. Suter, D. |
Citation: | Proceedings of the 14th Pacific-Asia Conference on Knowledge Discovery and Data Mining (PAKDD 2010), held in Hyderabad, India, 21-24 June 2010: pp.142-149 |
Publisher: | Springer-Verlag |
Publisher Place: | Germany |
Issue Date: | 2010 |
Series/Report no.: | Lecture notes in Computer Science ; 6118 |
ISBN: | 3642136567 9783642136573 |
ISSN: | 0302-9743 1611-3349 |
Conference Name: | Pacific-Asia Conference on Knowledge Discovery and Data Mining (14th : 2010 : Hyderabad, India) |
Editor: | Zaki, M.J. Yu, J.X. Ravindran, B. Pudi, V. |
Statement of Responsibility: | Nayyar Abbas Zaidi, David McG. Squire and David Suter |
Abstract: | A Nearest Neighbor (NN) classifier assumes class conditional probabilities to be locally smooth. This assumption is often invalid in high dimensions and significant bias can be introduced when using the nearest neighbor rule. This effect can be mitigated to some extent by using a locally adaptive metric. In this work we propose an adaptive metric learning algorithm that learns an optimal metric at the query point. We learn a distance metric using a feature relevance measure inspired by boosting. The modified metric results in a smooth neighborhood that leads to better classification results. We tested our technique on major UCI machine learning databases and compared the results to state of the art techniques. Our method resulted in significant improvements in the performance of the K-NN classifier and also performed better than other techniques on major databases. |
Keywords: | Adaptive Metric Learning Nearest Neighbor Bias-Variance analysis Curse-of-Dimensionality Feature Relevance Index |
Rights: | Copyright Springer-Verlag Berlin Heidelberg 2010 |
DOI: | 10.1007/978-3-642-13657-3_17 |
Published version: | http://www.springerlink.com/content/978-3-642-13671-9/?k=suter |
Appears in Collections: | Aurora harvest 5 Computer Science publications |
Files in This Item:
There are no files associated with this item.
Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.