[ros-users] An open source machine learning library (C++)

[Yanzhen Wang]

Hi all,

Machine learning is becoming more and more important in robotics, and robots today are gaining more and more learning abilities. Typical machine learning tools commonly used in robotics community include openCV and Matlab. But they have their own limitations, in either performance or target problem size. We'd like to share an open source ensemble learning library, which we call libEDM, with the community. It is written in C++ and provides functions covering almost all facets of ensemble learning. It can also be used as a toolkit of typical supervised machine learning algorithms, including basic classifiers such as support vector machines (SVM), back-propagation neural networks (BPNN), C4.5 decision trees, naive Bayes, and much more. One of the most important feature of our libEDM is its ability of dealing with extremely large datasets. The largest training set we tested contains 39,859,272 records, each with 75 attributes.

If you want to find more about libEDM or try it out, please go to its GitHub repository:
https://github.com/Qiangli-Zhao/LibEDM

And we are in the position of ROSifying some parts of the library to make it more usable in the robotics context. Please stay posted. Any feedback will be greatly appreciated.

Best regards,
Yanzhen 		 	   		  
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.ros.org/pipermail/ros-users/attachments/20140918/a93e315b/attachment.html>