The NBNN kernel
Authors:
- 1 November 2011
- conference paper
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- No. 15505499,p. 1824-1831
Abstract
Naive Bayes Nearest Neighbor (NBNN) has recently been proposed as a powerful, non-parametric approach for object classification, that manages to achieve remarkably good results thanks to the avoidance of a vector quantization step and the use of image-to-class comparisons, yielding good generalization. In this paper, we introduce a kernelized version of NBNN. This way, we can learn the classifier in a discriminative setting. Moreover, it then becomes straightforward to combine it with other kernels. In particular, we show that our NBNN kernel is complementary to standard bag-of-features based kernels, focussing on local generalization as opposed to global image composition. By combining them, we achieve state-of-the-art results on Caltech101 and 15 Scenes datasets. As a side contribution, we also investigate how to speed up the NBNN computations.Keywords
This publication has 10 references indexed in Scilit:
- Locality-constrained Linear Coding for image classification2010
- Robust classification of objects, faces, and flowers using natural image statistics2010
- Multiple kernels for object detection2009
- In defense of Nearest-Neighbor based image classification2008
- Why is Real-World Visual Object Recognition Hard?PLOS Computational Biology, 2008
- Multiclass multiple kernel learning2007
- Local Ensemble Kernel Learning for Object Category Recognition2007
- Object retrieval with large vocabularies and fast spatial matching2007
- Support Kernel Machines for Object Recognition2007
- Recognition with local features: the kernel recipe2003