Python (Theano) implementation of Learning a Metric for Class-Conditional KNN code provided by Daniel Jiwoong Im and Graham W Taylor.
Class Conditiaonl Metri Learning (CCML) learn a metric which captures perceptual similarity. Similar to how Neighbourhood Components Analysis optimizes a differentiable form of KNN classification, which optimizes a soft form of the Naive Bayes Nearest Neighbour (NBNN) selection rule. For more information, see
@article{Im2016ccml,
title={Learning a Metric for Class-Conditional KNN},
author={Im, Daniel Jiwoong and Taylor, Graham W.},
journal={International Joint Conference on Neural Networks (To appear)},
year={2016}
}If you use this in your research, we kindly ask that you cite the above workshop paper
Packages
Entry code for one-bit flip and factored minimum probability flow for mnist data are
- /test_ccml2_mnist.py
- /test_ccml2_norb.py