As others have said, you need to at least get the gist of what's going on mathematically to be able to make sensible decisions about model selection, structure,

As others have said, you need to at least get the gist of what's going on mathematically to be able to make sensible decisions about model selection, structure, etc. That said, I find a lot of the introductions to the theory behind ML techniques to be very poorly written. It's often worth giving a new student a conceptual simplification before introducing a rigorous definition. Without linear algebra and basic probability/calculus though, forget it. Luckily there's great sources to brush up on it.