mLMNN2.3
所属分类:matlab编程
开发工具:matlab
文件大小:3837KB
下载次数:65
上传日期:2013-11-30 17:48:16
上 传 者:
lalahong
说明: 多核处理多特征的融合(matlab实现——提供测试数据)
(Multicore processing multi-feature fusion (matlab achieve- to provide test data))
文件列表:
mLMNN2.3 (0, 2012-11-14)
mLMNN2.3\.svn (0, 2012-11-01)
mLMNN2.3\.svn\entries (169, 2012-11-01)
mLMNN2.3\.svn\format (2, 2012-11-01)
mLMNN2.3\.svn\prop-base (0, 2012-11-01)
mLMNN2.3\.svn\props (0, 2012-11-01)
mLMNN2.3\.svn\text-base (0, 2012-11-01)
mLMNN2.3\.svn\tmp (0, 2012-11-01)
mLMNN2.3\.svn\tmp\prop-base (0, 2012-11-01)
mLMNN2.3\.svn\tmp\props (0, 2012-11-01)
mLMNN2.3\.svn\tmp\text-base (0, 2012-11-01)
mLMNN2.3\applypca.m (526, 2012-11-01)
mLMNN2.3\data (0, 2012-11-01)
mLMNN2.3\data\.svn (0, 2012-11-01)
mLMNN2.3\data\.svn\all-wcprops (329, 2012-11-01)
mLMNN2.3\data\.svn\dir-prop-base (55, 2012-11-01)
mLMNN2.3\data\.svn\entries (527, 2012-11-01)
mLMNN2.3\data\.svn\format (2, 2012-11-01)
mLMNN2.3\data\.svn\prop-base (0, 2012-11-01)
mLMNN2.3\data\.svn\prop-base\bal2.mat.svn-base (80, 2012-11-01)
mLMNN2.3\data\.svn\prop-base\iris2.mat.svn-base (80, 2012-11-01)
mLMNN2.3\data\.svn\props (0, 2012-11-01)
mLMNN2.3\data\.svn\text-base (0, 2012-11-01)
mLMNN2.3\data\.svn\text-base\bal2.mat.svn-base (21168, 2012-11-01)
mLMNN2.3\data\.svn\text-base\iris2.mat.svn-base (5488, 2012-11-01)
mLMNN2.3\data\.svn\tmp (0, 2012-11-01)
mLMNN2.3\data\.svn\tmp\prop-base (0, 2012-11-01)
mLMNN2.3\data\.svn\tmp\props (0, 2012-11-01)
mLMNN2.3\data\.svn\tmp\text-base (0, 2012-11-01)
mLMNN2.3\data\digits.mat (3592820, 2012-11-01)
mLMNN2.3\demo.m (2950, 2012-11-01)
__MACOSX (0, 2013-04-04)
__MACOSX\mLMNN2.3 (0, 2013-04-04)
__MACOSX\mLMNN2.3\._demo.m (82, 2012-11-01)
mLMNN2.3\helperfunctions (0, 2012-11-01)
mLMNN2.3\helperfunctions\.svn (0, 2012-11-01)
mLMNN2.3\helperfunctions\.svn\all-wcprops (1420, 2012-11-01)
mLMNN2.3\helperfunctions\.svn\entries (1604, 2012-11-01)
mLMNN2.3\helperfunctions\.svn\format (2, 2012-11-01)
mLMNN2.3\helperfunctions\.svn\prop-base (0, 2012-11-01)
... ...
*** Changelog
Update 13/11/2012
- Fixed a bug that prevented execution with even values for k.
Update 01/11/2012
- Added optional 'diagonal' version to learn diagonal matrices
Update 09/19/2012
- Added 32-bit Windows binaries (Thanks to Ya Shi)
Update 09/18/2012
- Added parameter 'outdim' to easily specify the output dimensionality
- Small fixes in mtree code, which broke compilation on some windows machines.
- Speedup in findimps3Dm by substituting some repmats with bsxfun (somehow they have been overlooked)
Update 09/13/2012
- Small fix to setpaths.m script
- Rearranged files to ensure that the mexed files are in the path.
- updated demo
Update 09/06/2012
- Small fix to instlall.m script
Update 08/23/2012
This package contains the implementation of Large Margin Nearest Neighbors (LMNN).
Changes from version 2.0 to 2.1:
- Removed mex files which are no longer faster than the Matlab equivalent (Matlab became a lot faster over the years)
- Updated mtrees to compile on windows computers and no longer use depreciated libraries
- Removed all BLAS / LAPACK dependencies
- Renamed knnclassify.m to knncl.m (as former clashed with the implementation from the statistics toolbox)
(Many thanks to Jake Gardner who helped a lot with tyding up of the code.)
*** Usage
Please run (inside the matlab console)
install
demo
to see a working demo.
If you use this code in scientific work, please cite:
@article{weinberger2009distance,
title={Distance metric learning for large margin nearest neighbor classification},
author={Weinberger, K.Q. and Saul, L.K.},
journal={The Journal of Machine Learning Research},
volume={10},
pages={207--244},
year={2009},
publisher={JMLR.org}
}
近期下载者:
相关文件:
收藏者: