Asymmetric bagging and random subspace for support vector machines-based relevance feedback in image retrieval
Authors
Discussion
Loading...
Member takes
Implication
Randomized trial reveals improved relevance feedback in image retrieval, indicating better performance with new methods.
Key Points
- The aim is to enhance support vector machines' performance in image retrieval through improved feedback methods.
- Developed asymmetric bagging-based SVM (AB-SVM) to stabilize classifiers with small training sets.
- Integrated random subspace technique with SVM for enhanced feature space management.
- Combined AB-SVM and random subspace SVM to create an innovative model (ABRS-SVM).
- ABRS-SVM significantly improved relevance feedback performance compared to traditional methods.
- Observed reductions in overfitting issues due to more balanced training data handling.
- Error rates decreased, particularly in scenarios with limited positive feedback samples.
Cite This Study
Tao et al. (2006) studied this question.