xichaoo / mrmr Goto Github PK

对mRMR的解析

Please run demo.m

• 代码进行了调整，删除了作者一些无效代码，看起来更加的清晰
• 作者使用了两种特征选取的方式，命名为mid 和miq

Mutual information computation, conditional probability and entropy estimation for discrete/categorical variables.

by Hanchuan Peng, 2007-May-3.

Address update: Note that the author's current email is [email protected] , please disregard the address info in the header part of each file.

License of using this package:

This package is free for non-profit use. For commercial use of this software, please contact the author, Hanchuan Peng.

Any use of this package should acknowledge the author, or the following paper:

Hanchuan Peng, Fuhui Long, and Chris Ding, "Feature selection based on mutual information: criteria of max-dependency, max-relevance, and min-redundancy," IEEE Transactions on Pattern Analysis and Machine Intelligence, Vol. 27, No. 8, pp.1226-1238, 2005.

This package cannot be re-distributed in any form, including the recompiled codes, without the explicit permission from the author, Hanchuan Peng. The author is NOT reliable to any damage in any form due to the use of this package.

Anybody that uses this package or changes the codes cannot distribute the products that include a part of this package (or the compiled form, or the renamed/revised and re-compiled forms) without the explicit permission from the original author, Hanchuan Peng.

Any inclusion or other use of this package means acceptance of this license.

2005-10-23: Fix a programming bug in estmutualinfo.cpp which might lead to segmentation fault is not input the p1 and p2 parameters. In fact this has not led to any dfference in the mutualinfo.m program, as I always call estmutualinfo using three parameters. BTW, I detected this bug when I wrote the C++ version of mRMR selection program.