I have already redefined the Internet, but I'm still not sure what the term "class labels" in the neural network means (classification) ...

It seems like this is a list of ALL output neurons (responses): (for example, if a neuron distinguishes a cat from a dog, it will have only two class marks: a cat and a dog (or more precisely, the numbers on which it is determined)), or are these just the right answers for learning neural network (for example, given several images for the same neuron, and they have written in the title, it is a cat or a dog).

It turned out confusing, but I hope it is clear.

I would be very grateful for the answer!

    1 answer 1

    Class labels are exactly class names. For an example with cats and dogs, this is a data column or a list consisting of these two values ​​( кот or собака ).

    The neural network feeds data in numerical form. To do this, usually use different encoders . As a result, you can get one column with numbers corresponding to unique classes or One-Hot Encoding .

    • Here, just wanted to ask about it. Thank you very much! - Nephilim