Home

Nak is a machine learning library written in Scala and Java. It currently supports the following:

• basic k-means clustering for small datasets
• classification with logistic regression and support vector machines. (The implementations are adapted pretty much directly from liblinear-java.)

Much is in flux with Nak. For example:

• We are currently pulling out the legacy opennlp maxent classes.
• We are merging in code from Breeze---essentially Nak will become the machine learning library of ScalaNLP in general. See Breeze Issue 51.
• Documentation is forthcoming.

Basically, Nak is undergoing a phase of contraction to be followed by expansion, in line with the overall ScalaNLP roadmap. Help with this process is welcome -- get in touch with Jason Baldridge if you are interested in contributing.