Linear and Non-Linear Models

[gents98]

Hello there,

A question popped up in my head while I was watching the visualization for multiclass-model when comparing the plots of using non-linear and linear regression models (the part where ReLU was commented off).

During the intro to binary classification, the model wasn't able to learn because the linear line couldnt seperate the reds from the blue, thus a non-linear model was needed. (See first picture)

However, for the multiclass-model plot below

the green line can be considered linear but the red line (even though its straight, it broke off halfway through) is also considered linear? I would suppose that linear lines would follow through all the way, like in the first picture. The reason for this question was because I recalled the graph function of ReLU which has 2 straights line which is considered non-linear which looks like the line in red.

So my question would be why is the multi-class linear model (without ReLU) considered linear when it doesnt exactly use linear lines to seperate the datas?

Im sorry if this question sounds confusing, im really bad at expressing my thoughts to words D:
Thanks for taking your time to read!