Help understanding machine learning video

53 Views Asked by At

I am currently following a long this video, which is an introduction to Machine Learning. (https://www.youtube.com/watch?v=esTIhqAFKu4&list=PLGd9Gn0_Oc65os52iy4jwvex2q50r5q0U&index=27&t=2818s)

At 47:15 to 48:45 the professor goes on to talk about, how "non normalizing" (sorry, I have a hard time hearing him even with subtitles on its a mess) the dimension of the vector goes down by 1. So having a vector in the realm of R^(p) gives a dimension of R^(p-1). He even uses the example of if a vector was in R^2 than it would be R^1. This seems wrong, but it might be me who have a hard time understanding him. Can someone brighter than me shine some lights on this? Thank you very much.

1

There are 1 best solutions below

0
On BEST ANSWER

The lecturer is saying "norm normalized", where "norm" refers to the usual Euclidean length/norm function written $|| \cdot ||$. If $x \in \mathbb{R}^p$ is nonzero then if we "norm normalize" it the result is $x / ||x||$. That's got norm 1, so it belongs to the $(p-1)$ sphere $S^{p-1}$ (because this is defined to be the set of vectors in $\mathbb{R}^p$ with norm 1).

For example, if you take a nonzero vector $x \in \mathbb{R}^2$ then normalize it you get something on the unit circle $S^1$.