Thanks for the post. To me, it is refreshing to see some ML in Java, aside from weka.
I am following closely what is happenning in GPU backed convnets at the moment. Do you plan on adding support of convnets in the near future? (wether it is mappings to cuda-convnet or NVIDIA's cuDNN, or even reimplementing it with jocl or jcuda)
Hi! Thanks for your message. It was not within my plans to add the support of convnets as I have no experience with GPU programming. Nevertheless now that it is open-sourced, if someone is interested in participating this feature could be supported. :)