Is Julia the new language of choice for machine learning?

Julia is one of the fastest growing programming languages of 2018. Flux a machine learning library written in Julia stack already makes it easier to write ML code.

Today, Python and R dominate machine learning and are continuing to grow too. How does Julia eco-system feel like to you? Is it going to be the new language of choice for ML?

Comments (1)

Mark's photo

Well, it's hard to overcome the kind of inertia that Python and R have. So I think it can go either way.

(Imho that inertia is the whole reason it's still Python and R, rather than just Python).

But I think that if some language disrupts the status quo in the field, it'll be probably be Julia. I think it has many of the important qualities

  • Quick prototyping, with some scalability (e.g. optional types)
  • Good performance (compared to py/R), and C interop
  • Easy for people for whom programming is a tool, not their main career
  • Focused on numerical applications

except the most important one

  • A massive ML ecosystem

Honestly, I don't think the difference is big enough for the switch to be fast, but I think new users who haven't already invested a lot in py/R may prefer it and cause it to grow.

Jason Knight posted a question earlier this year, perhaps he can say smarter things about it than me.