I used Clojure to play with word vectors through Deep Learning for Java. I wrote up a walkthrough to read, or to play with in your REPL. The article touches on Java interop, data visualization, machine learning, refactoring, and much more.
- Now I learned that
Word2Vec$Builderin a Clojure import refers to the
Builderclass embedded in the
- Your REPL use really helps bring the more abstract concepts down to earth.
- And really interesting to read about “subtracting” words. The way you write it, it makes sense, in a weird way.
- And overall I had a few good laughs reading the “X is to Y as Z is to _” examples you’re going through.
I was a little surprised to see a “deep learning toolbox specialized for word analysis”. But when I read more about it, it really makes sense. Do you think that’s something we’ll see more of for other applications?