In a nutshell

Optimal Naive Bayes Nearest Neighbors (oNBNN) is a machine learning algorithm for the classification of objects that come under the form of one or multiple sets of multi-dimensional points. It was designed to be applied to automated image classification, but can also be employed in a variety of contexts, such as text and sound classification.

We provide here a C++ implementation of the main bricks of the algorithm.

References

• oNBNN: Towards Optimal Naive Bayes Nearest Neighbors, R. Behmo, P. Marcombes, A. Dalalyan & V. Prinet, proceedings of the 2010 European Conference on Computer Vision (ECCV 2010), available here.
• NBNN: our classifier was initally based on Naive Bayes Nearest Neighbours (NBNN), which was described in: In Defense of Nearest-Neighbor Based Image Classification, O. Boiman, E. Shechtman, M. Irani, available here.

If you decide to make use of our code for research work, please use the following bibtex reference:

@inproceedings{optimal_NBNN_ECCV2010,
  author = {Behmo, Régis and Marcombes, Paul and Dalalyan, Arnak and Prinet, Véronique},
  title = {Towards Optimal Naive Bayes Nearest Neighbors},
  booktitle = {European Conference on Computer Vision (ECCV)},
  year = {2010}
}

Acknowledgements

This research work is the result of a collaboration between the following authors:

• Régis Behmo, Ecole Centrale Paris - Chinese Academy of Sciences
• Paul Marcombes, Ecole des Ponts ParisTech - Chinese Academy of Sciences
• Arnak Dalalyan, Ecole des Ponts ParisTech
• Véronique Prinet, Chinese Academy of Sciences

As such, this work was supported by several grants and institutions, including: the French National Institute for Automation and Computer Science (INRIA) and the Chinese Ministry of Science and Technology.