3 ms·
> Machine learning is pretty much all math. Eh, I would say math is the language we use to communicate to computers presently, but the underlying _concepts_ do
by tomrod 9y ago
> Machine learning is pretty much all math.
Eh, I would say math is the language we use to communicate to computers presently, but the underlying _concepts_ don't require "more maths" and can often be grasped through an intuitive approach. For example, almost everyone who uses photoshop understands the underlying concept of a convolution (e.g. Gaussian blur), even if they don't know the mathematics that can be used to describe the operation. Yes, there are difficult notions that formalization or generalization assist with--perhaps it is better to see math as augmenting the initial intuition, rather than driving the intuition?
- heavenlyblue 9y agoIt's not always helpful even when someone gives you a correct direction but none of his abstracted vision of how to apply that direction. For example: convolution is not _just_ Gaussian blur; it allows someone to find an object on the scene (or a shape of something in time dimension). How is that related to Gaussian blur and why are they the same? It takes time to understand the full domain of the concept.
- tomrod 9y agoTo understand the underlying principles, but not to use it. In this case, it's a point and click -- if they like the effect, keep it. Literally no maths are required for understanding, yet the outcome is achieved.