Column subset selection for active learning in image classification

Authors:
Jianfeng Shen;Bin Ju;Tao Jiang;Jingjing Ren;Miao Zheng;Chengwei Yao;Lanjuan Li
Affiliations:
The First Affiliated Hospital of College of Medical School, Zhejiang University, China and Zhejiang Medical College, China and Zhejiang Province Health Bureau Center of Information, China;Zhejiang Province Health Bureau Center of Information, China;Zhejiang Province Health Bureau Center of Information, China;The First Affiliated Hospital of College of Medical School, Zhejiang University, China;College of Computer Science and Technology, Zhejiang University, China;College of Computer Science and Technology, Zhejiang University, China;State Key Laboratory for Diagnosis and Treatment of Infectious Diseases, No. 216, Qingchun Road, Hangzhou, Zhejiang 310006, China
Venue:
Neurocomputing
Year:
2011

Citing 26
Cited 0

Some applications of the rank revealing QR factorization

SIAM Journal on Scientific and Statistical Computing
The nature of statistical learning theory

The nature of statistical learning theory
Computing rank-revealing QR factorizations of dense matrices

ACM Transactions on Mathematical Software (TOMS)
Nonlinear component analysis as a kernel eigenvalue problem

Neural Computation
An introduction to support Vector Machines: and other kernel-based learning methods

An introduction to support Vector Machines: and other kernel-based learning methods
Support vector machine active learning for image retrieval

MULTIMEDIA '01 Proceedings of the ninth ACM international conference on Multimedia
Kernel-Based Object Tracking

IEEE Transactions on Pattern Analysis and Machine Intelligence
Support vector machine active learning with applications to text classification

The Journal of Machine Learning Research
A Semi-Supervised Active Learning Framework for Image Retrieval

CVPR '05 Proceedings of the 2005 IEEE Computer Society Conference on Computer Vision and Pattern Recognition (CVPR'05) - Volume 2 - Volume 02
Active Learning with Feedback on Features and Instances

The Journal of Machine Learning Research
Laplacian optimal design for image retrieval

SIGIR '07 Proceedings of the 30th annual international ACM SIGIR conference on Research and development in information retrieval
Pool-based active learning with optimal sampling distribution and its information geometrical interpretation

Neurocomputing
Error-driven active learning in growing radial basis function networks for early robot learning

Neurocomputing
Hierarchical sampling for active learning

Proceedings of the 25th international conference on Machine learning
Semi-supervised and active learning with the probabilistic RBF classifier

Neurocomputing
Unsupervised feature selection for principal components analysis

Proceedings of the 14th ACM SIGKDD international conference on Knowledge discovery and data mining
Geometric Mean for Subspace Selection

IEEE Transactions on Pattern Analysis and Machine Intelligence
Column subset selection, matrix factorization, and eigenvalue optimization

SODA '09 Proceedings of the twentieth Annual ACM-SIAM Symposium on Discrete Algorithms
Uncertainty sampling and transductive experimental design for active dual supervision

ICML '09 Proceedings of the 26th Annual International Conference on Machine Learning
Active learning with statistical models

Journal of Artificial Intelligence Research
Convex experimental design using manifold structure for image retrieval

MM '09 Proceedings of the 17th ACM international conference on Multimedia
A general kernelization framework for learning algorithms based on kernel PCA

Neurocomputing
Kernel methods and the exponential family

Neurocomputing
Max-Min Distance Analysis by Using Sequential SDP Relaxation for Dimension Reduction

IEEE Transactions on Pattern Analysis and Machine Intelligence
Manifold elastic net: a unified framework for sparse dimension reduction

Data Mining and Knowledge Discovery
Active Learning Based on Locally Linear Reconstruction

IEEE Transactions on Pattern Analysis and Machine Intelligence

Quantified Score

Hi-index	0.01

Visualization

Abstract

Image classification is an important task in computer vision and machine learning. However, it is known that manually labeling images is time-consuming and expensive, but the unlabeled images are easily available. Active learning is a mechanism which tries to determine which unlabeled data points would be the most informative (i.e., improve the classifier the most) if they are labeled and used as training samples. In this paper, we introduce the idea of column subset selection, which aims to select the most representation columns from a data matrix, into active learning and propose a novel active learning algorithm, column subset selection for active learning (CSS"a"c"t"i"v"e). CSS"a"c"t"i"v"e selects the most representative images to label, then the other images are reconstructed by these labeled images. The goal of CSS"a"c"t"i"v"e is to minimize the reconstruction error. Besides, most of the previous active learning approaches are based on linear model, and hence they only consider linear functions. Therefore, they fail to discover the intrinsic geometry in images when the image space is highly nonlinear. Therefore, we provide a kernel-based column subset selection for active learning (KCSS"a"c"t"i"v"e) algorithm which performs the active learning in Reproducing Kernel Hilbert Space (RKHS) instead of the original image space to address this problem. Experimental results on Yale, AT&T and COIL20 data sets demonstrate the effectiveness of our proposed approaches.