Machine learning algorithms, at least the ones that I used, needs convex optimization to fully understand them which needs real analysis which is just calculus with more mathematical rigor.