Change search
ReferencesLink to record
Permanent link

Direct link
Improved boosting performance by explicit handling of ambiguous positive examples
KTH, School of Computer Science and Communication (CSC), Computer Vision and Active Perception, CVAP.
KTH, School of Computer Science and Communication (CSC), Computer Vision and Active Perception, CVAP.
2015 (English)In: Pattern Recognition: Applications and Methods, Springer Berlin/Heidelberg, 2015, 17-37 p.Conference paper (Refereed)
Abstract [en]

Visual classes naturally have ambiguous examples, that are different depending on feature and classifier and are hard to disambiguate from surrounding negatives without overfitting. Boosting in particular tends to overfit to such hard and ambiguous examples, due to its flexibility and typically aggressive loss functions. We propose a two-pass learning method for identifying ambiguous examples and relearning, either subjecting them to an exclusion function or using them in a later stage of an inverted cascade. We provide an experimental comparison of different boosting algorithms on the VOC2007 dataset, training them with and without our proposed extension. Using our exclusion extension improves the performance of almost all of the tested boosting algorithms, without adding any additional test-time cost. Our proposed inverted cascade adds some test-time cost but gives additional improvements in performance. Our results also suggest that outlier exclusion is complementary to positive jittering and hard negative mining.

Place, publisher, year, edition, pages
Springer Berlin/Heidelberg, 2015. 17-37 p.
, Advances in Intelligent Systems and Computing, ISSN 2194-5357 ; 318
Keyword [en]
Algorithm evaluation, Boosting, Dataset pruning, Image classification, VOC2007, Pattern recognition, Boosting algorithm, Experimental comparison, Outlier exclusion, Positive examples, Classification (of information)
National Category
Computer and Information Science
URN: urn:nbn:se:kth:diva-167371DOI: 10.1007/978-3-319-12610-4_2ISI: 000364822300002ScopusID: 2-s2.0-84914145602ISBN: 9783319126098OAI: diva2:815524
nd International Conference on Pattern Recognition, ICPRAM 2013; Barcelona; Spain; 15 February 2013 through 18 February 2013

QC 20150601

Available from: 2015-06-01 Created: 2015-05-22 Last updated: 2015-12-17Bibliographically approved

Open Access in DiVA

No full text

Other links

Publisher's full textScopus

Search in DiVA

By author/editor
Kobetski, MiroslavSullivan, Josephine
By organisation
Computer Vision and Active Perception, CVAP
Computer and Information Science

Search outside of DiVA

GoogleGoogle Scholar
The number of downloads is the sum of all downloads of full texts. It may include eg previous versions that are now no longer available

Altmetric score

Total: 80 hits
ReferencesLink to record
Permanent link

Direct link