A Caltech Library Service

Pruning training sets for learning of object categories

Angelova, Anelia and Abu-Mostafa, Yaser S. and Perona, Pietro (2005) Pruning training sets for learning of object categories. In: Conference on Computer Vision and Pattern Recognition (CVPR '05),San Diego, CA, 20-25 June 2005. Vol.1. IEEE , Piscataway, NJ, pp. 494-501. ISBN 0-7695-2372-2.

PDF - Published Version
See Usage Policy.


Use this Persistent URL to link to this item:


Training datasets for learning of object categories are often contaminated or imperfect. We explore an approach to automatically identify examples that are noisy or troublesome for learning and exclude them from the training set. The problem is relevant to learning in semi-supervised or unsupervised setting, as well as to learning when the training data is contaminated with wrongly labeled examples or when correctly labeled, but hard to learn examples, are present. We propose a fully automatic mechanism for noise cleaning, called ’data pruning’, and demonstrate its success on learning of human faces. It is not assumed that the data or the noise can be modeled or that additional training examples are available. Our experiments show that data pruning can improve on generalization performance for algorithms with various robustness to noise. It outperforms methods with regularization properties and is superior to commonly applied aggregation methods, such as bagging.

Item Type:Book Section
Related URLs:
URLURL TypeDescription
Perona, Pietro0000-0002-7583-5809
Additional Information:© 2005 IEEE. Reprinted with permission. Publication Date: 20-25 June 2005. Date Published in Issue: 2005-07-25. This research is supported by the NSF Center for Neuromorphic Systems Engineering grant EEC-9402726.
Funding AgencyGrant Number
Center for Neuromorphic Systems Engineering, CaltechUNSPECIFIED
Subject Keywords:face recognition; machine learning; object detection; aggregation method; data pruning; noise cleaning; object category learning; training set pruning
Record Number:CaltechAUTHORS:ANGcvpr05
Persistent URL:
Usage Policy:No commercial reproduction, distribution, display or performance rights in this work are provided.
ID Code:11469
Deposited By: Kristin Buxton
Deposited On:08 Sep 2008 18:15
Last Modified:03 Oct 2019 00:19

Repository Staff Only: item control page