Distinctive Image Features from Scale-Invariant Keypoints
International Journal of Computer Vision
Mining Ratio Rules Via Principal Sparse Non-Negative Matrix Factorization
ICDM '04 Proceedings of the Fourth IEEE International Conference on Data Mining
A Bayesian Hierarchical Model for Learning Natural Scene Categories
CVPR '05 Proceedings of the 2005 IEEE Computer Society Conference on Computer Vision and Pattern Recognition (CVPR'05) - Volume 2 - Volume 02
Multi-label informed latent semantic indexing
Proceedings of the 28th annual international ACM SIGIR conference on Research and development in information retrieval
Multi-labelled classification using maximum entropy method
Proceedings of the 28th annual international ACM SIGIR conference on Research and development in information retrieval
MILES: Multiple-Instance Learning via Embedded Instance Selection
IEEE Transactions on Pattern Analysis and Machine Intelligence
International Journal of Computer Vision
Correlative multi-label video annotation
Proceedings of the 15th international conference on Multimedia
Exploiting spatial context constraints for automatic image region annotation
Proceedings of the 15th international conference on Multimedia
Semi-supervised multi-label learning by constrained non-negative matrix factorization
AAAI'06 Proceedings of the 21st national conference on Artificial intelligence - Volume 1
NUS-WIDE: a real-world web image database from National University of Singapore
Proceedings of the ACM International Conference on Image and Video Retrieval
Unified Solution to Nonnegative Data Factorization Problems
ICDM '09 Proceedings of the 2009 Ninth IEEE International Conference on Data Mining
A Bayesian network-based framework for semantic image understanding
Pattern Recognition
ECCV'06 Proceedings of the 9th European conference on Computer Vision - Volume Part I
Mining Multilevel Image Semantics via Hierarchical Classification
IEEE Transactions on Multimedia
IEEE Transactions on Image Processing
Hi-index | 0.00 |
Conventional semi-supervised learning algorithms over multi-label image data propagate labels predominantly via the holistic image similarities, ignoring that each label essentially only characterizes a local region within an image. In this paper, we present a novel propagation-by-decomposition solution to this problem with the following characteristics: 1) each image representation is implicitly decomposed into several label representations; 2) those decompositions are guided by the so-called hidden concepts, which are expected to characterize image regions and be able to reconstruct both visual and non-visual labels of the entire image label space; 3) the intra-label diversity is expressed by the hidden-concept-specific subspace, which acts as the intermediate entity for propagating specific label from labeled data to unlabeled ones; and 4) the sparse coding based graph is proposed to enforce the collective consistency between image labels and image representations, which naturally avoids the dilemma of possible inconsistency between the pairwise label similarity and image representation similarity in multi-label scenario. These properties are finally embodied in a regularized nonnegative data factorization formulation, from which a convergence provable updating procedure is presented to iteratively optimize the objective function. Extensive experiments on three benchmark image datasets well validate the effectiveness of our proposed solution to semi-supervised multi-label image annotation problem.