search

airoldilab / sgd

An R package for large scale estimation with stochastic gradient descent
60 stars 18 forks source link

sparse matrices support? #85

Open dselivanov opened 8 years ago

dselivanov commented 8 years ago

I want to try sgd package - seems it provides a lot of options and must-have features. But why it didn't work with sparse matrices (Matrix package, especially dgCMatrix class, as it default for R usage)? Working with sparse matrices is a core functionality for working with any large-scale problems. Or mb I missed something?

dustinvtran commented 8 years ago

Good question! The only reason is because it hasn't been implemented yet. :) For the current work, we were only considering dense matrices for various simulation studies; of course, in practice it would be incredibly helpful to get sparse matrix support. I'd like to work on this eventually. Feel free to submit a pull request if you'd like to yourself.

dselivanov commented 8 years ago

Thank you for quick response, I'll happy to contribute in future. PS Also I found your blog very interesting!

ptoulis commented 8 years ago

This is a great point. Eventually we will work on supporting sparse data formats is a priority, as well as parallelization.