Hacker News new | ask | show | jobs
Mlpack: A C++ machine learning library (github.com)
47 points by ajinkyakale 4198 days ago
3 comments

The project is currently in a state of transition to Github from svn, which explains the lack of issues and presence. http://www.mlpack.org/ has links to all of the old development tools such as Trac (http://www.mlpack.org/trac/) which will be ported to Github soon.
Has anyone actively used this for any project (big or small)? Looks like the project has 10K+ commits, so we can reasonably assume it's mature.
Development of mlpack has been going on for at least 7 years and my impression is that it's pretty mature. It was originally affiliated with a lab at Georgia Tech, where its current maintainer (a friend of mine) is a PhD student. The dual-tree methods are based on his research.
Here's a list of some (academic) uses of mlpack:

http://scholar.google.ca/scholar?cites=17794023163287158191&...

I've used it, it's great. My main interest was using some of the regularization methods, which all worked as expected.

I was able to talk to some of the maintainers on IRC, very responsive and tolerant of my stupid questions.

Why is the project relying on boost::random, instead of using std::random? Is there a specifically reason for such a choice?