Is functional programming necessary in machine learning?

Question

Haskell

I'm interested in areas somewhat related to AI, in particular, computer vision and automatic word processing.
There was an opportunity to attend the course "Functional Programming", where the program studies the Lisp dialect (scheme) smoothly turning into Haskell. I have nothing against functional languages ​​and seem to understand the differences from imperative programming languages, however, I don’t understand where this very functional programming is applicable, and whether it is needed, for example, in machine learning.
Of course, for the overall development of FP in any case it is useful, but in fact you will have to spend a lot of time on the course (and travel). That’s why the question is: maybe it’s worth spending time on a more extensive study of the theory of machine learning, rather than engaging in functionalism? Googling it seemed that it is very few where applicable in real life.

Answer 1

For the general development - it is necessary.
Specifically for AI, there is no direct connection with functionality.

Answer 2

For example, the great openCV, which uses many machine learning algorithms, was originally written in C, then C ++ and comrades joined it. Need of course. In pure SI, you can write almost everything and even more. By the way, si is trending in 2(!!!) place, overtaking almost everyone. Draw your own conclusions. link here .

Answer 3

Now ML is mostly done in languages ​​with poor functional style support.
But many things from ML theory are usually stated in terms of pure functions, functional programming gives a new look at working with containers and streams, frameworks for working with big data are to some extent functionally oriented. So there is a chance that the proportion of languages ​​with FP in ML will grow - Scala and Julia are gaining popularity, and R is still holding on.
Just written