Ensembles of decision trees (boosted, bagged) in golang. It's a project I started as a research platform and it has some great features not found elsewhere and good computational performance. There are also a couple of companies using it in production and some researchers using it in computational genetics.
I'm just starting development again after a bit of a hyatus. This month my main goal is to do some basic maintenance and work with some people who've forked it to bring features and bug fixes back into main.
Then I want to duplicate some xgboost like features including parallelism that will work with boosting and 2nd order gradient boosting (using the hessian instead of just the gradient).
And i'd like to to utilize go's newish shared object support to make it callable from python and R.
Skills Needed: go programmers confident in writing low level machine learning style code or python/R/etc programmers interested in maintaining a wrapper. Data scientists interested in using it and providing feedback on performance.
Comments
CloudForest
Ensembles of decision trees (boosted, bagged) in golang. It's a project I started as a research platform and it has some great features not found elsewhere and good computational performance. There are also a couple of companies using it in production and some researchers using it in computational genetics.
https://github.com/ryanbressler/CloudForest
BSD-3-clause
This month:
I'm just starting development again after a bit of a hyatus. This month my main goal is to do some basic maintenance and work with some people who've forked it to bring features and bug fixes back into main.
Then I want to duplicate some xgboost like features including parallelism that will work with boosting and 2nd order gradient boosting (using the hessian instead of just the gradient).
And i'd like to to utilize go's newish shared object support to make it callable from python and R.
Skills Needed: go programmers confident in writing low level machine learning style code or python/R/etc programmers interested in maintaining a wrapper. Data scientists interested in using it and providing feedback on performance.
Onboarding: I just set up a google group where people can introduce themselves: https://groups.google.com/forum/#!forum/cloudforest-dev
I'll work on cleaning out the issue tracker and prioritizing things so people can pick them up if they want.