These two libraries form another set of task-specific examples of ML libraries in Go. Both are well-written and offer primitives specific to their features, but these primitives do not generalize. In the Naive Bayes classifier lib, matrices are built manually before they can be used, while the traditional approach to generic algorithms makes no use of matrices at all. There has been some work on incorporating them into GA; however, this work is yet to make it into the GA library we're referencing here.