How a line could separate samples into 3 classes

by zorkor   Last Updated July 12, 2018 01:19 AM

I have troubling the understand linear separability issue. As far as I search on the web,

To be linearly separable, you have to able to separate data with a straight line for 2D, plane for 3D,

From that sentence, I understand that, I have to separate data with a single line if it is linearly separable. If I'm right, I wonder how could I separate my data into 3 different classes with a single line? It seems impossible?

Answers 1

It's just a definition. If you can separate classes with a straight line (in 2 dimensions), the data are linearly separable.

If you have 3 classes in 2 dimensions, they are linearly separable if you can separate each class from all others singly, like this

a | b | c

or something similar. (This is sometimes called the "one versus all" strategy.)

July 12, 2018 01:14 AM

Related Questions

Difference between "kernel" and "filter" in CNN

Updated August 31, 2018 05:19 AM

What does the term saturating nonlinearities mean?

Updated October 10, 2017 11:19 AM