Making image to class distance comparable

Authors:
Deyuan Zhang;Bingquan Liu;Chengjie Sun;Xiaolong Wang
Affiliations:
School of Computer Science and Technology, Harbin Institute of Technology, China;School of Computer Science and Technology, Harbin Institute of Technology, China;School of Computer Science and Technology, Harbin Institute of Technology, China;School of Computer Science and Technology, Harbin Institute of Technology, China
Venue:
ICONIP'11 Proceedings of the 18th international conference on Neural Information Processing - Volume Part II
Year:
2011

Citing 10
Cited 0

Support-Vector Networks

Machine Learning
Pattern Classification (2nd Edition)

Pattern Classification (2nd Edition)
Distinctive Image Features from Scale-Invariant Keypoints

International Journal of Computer Vision
Generic Object Recognition with Boosting

IEEE Transactions on Pattern Analysis and Machine Intelligence
One-Shot Learning of Object Categories

IEEE Transactions on Pattern Analysis and Machine Intelligence
An introduction to ROC analysis

Pattern Recognition Letters - Special issue: ROC analysis in pattern recognition
Performance measures for multilabel evaluation: a case study in the area of image classification

Proceedings of the international conference on Multimedia information retrieval
Image-to-class distance metric learning for image classification

ECCV'10 Proceedings of the 11th European conference on Computer vision: Part I
Towards optimal naive bayes nearest neighbor

ECCV'10 Proceedings of the 11th European conference on Computer vision: Part IV
Face and Human Gait Recognition Using Image-to-Class Distance

IEEE Transactions on Circuits and Systems for Video Technology

Quantified Score

Hi-index	0.00

Visualization

Abstract

Image classification is to classify the image into predefined image categories. The image to class distance(I2CD), with simple formulation, can tackle the intra-class variation and show the state of the art results in several datasets. This paper focuses on the performance of I2CD on imbalanced training dataset which has not been catched much attention by I2CD researchers. Under the naive bayes assumption, when the dataset is imbalanced, I2CD is not comparable. We propose Random Sampling I2CD to tackle the imbalanced problem, and provide an efficient approximation method to reduce the test time complexity. Experimental results show that PRSI2CD outperforms the original I2CD in imbalanced setting.