We began by developing a semi-supervised learning method based on the expectation-maximization (EM) algorithm, and maximum likelihood and maximum a posteriori classifiers (MLC and MAP). This scheme utilizes a small set of labeled and a large number of unlabeled training samples. We conducted several experiments on multi-spectral images to understand the impact of unlabeled samples on the classification performance. Our study shows that although, in general, classification accuracy improves with the addition of unlabeled training samples, it is not guaranteed to achieve consistently higher accuracies unless sufficient care is exercised when designing a semi-supervised classifier. We also extended this semi-supervised framework to model spatial context through Markov random fields (MRF). Initial experiments showed an improved accuracy of the spatial semi-supervised algorithm (SSSL) over MLC, semi-supervised, and MRF classifiers. An efficient implementation is provided so that the SSSL can be applied in production environments. We also discuss some open research problems.
|Original language||English (US)|
|Number of pages||11|
|Journal||International Journal of Parallel, Emergent and Distributed Systems|
|State||Published - Jan 2007|
Bibliographical noteFunding Information:
This research has been supported in part by the Army High Performance Computing Research Center under the auspices of Department of the Army, Army Research Laboratory Cooperative agreement number DAAD19-01-2-0014, and by the cooperative agreement with NASA (NCC 5316) and by the University of Minnesota Agriculture Experiment Station project MIN-42-044. We are particularly grateful to our collaborator Prof. Joydeep Ghosh for useful conversations and critical inputs. We are thankful to Tim Mack, who collected ground truth data for this study. We greatly benefited from discussions with researchers at the Remote Sensing lab and Spatial Database research group. Reviewers comments from the KDD/MDM 2006 workshop have also helped us in improving the technical quality of this paper. We would like to thank Kim Koffolt for improving the readability of this paper.
- Random fields; image analysis
- Semi-supervised learning