This code is related to the distance-based implementation of Large Margin Nearest Neighbor classifier (dist-LMNN)
- Run install.m file first to prepared the mex-files and set the directory paths.
- The dist_LMNN_demo.m contains a test run of the dist-LMNN algorithm on 3 databases with pre-calculated distance matrices.
- In order to use dist_LMNN on another dataset, please prepare its pair-wise distance matrix (via any preferred distance) as instructed in the main m-file.
Hosseini, Babak, and Barbara Hammer. "Efficient metric learning for the analysis of motion data", IEEE International Conference on Data Science and Advanced Analytics (DSAA). IEEE, 2015.