International Conference on Instrumentation, Measurement, Circuits and Systems (ICIMCS 2011)
179 A New Multi-Instance Learning Scheme for Scene Categorization Using Information Bottleneck Theory
Download citation file:
- Ris (Zotero)
- Reference Manager
Multi-Instance learning (MIL) as a learning framework proposed recently has been successfully used in scene and video classification. This paper first proposes a new image Multi-Instance (MI) bag generating method, which models an image with a Gaussian Mixed Model (GMM). The generated GMM is treated as an MI bag, of which the color and locally stable invariant components (SIFT) are the instances. Then, hierarchical clustering is employed to transform the MIL problem into single-instance learning problem so that single-instance classifiers can be used for classification. Finally, ensemble learning is involved to further enhance classifiers generalization ability. Experimental results demonstrate that the performance of the proposed framework for image classification is superior to some common MI algorithms on average in a 5-class scene classification task.