How to build a training dataset for process processing?

Question

Neural networks

Hello! There is a signal. How best to choose a training set from it?
1 Take a section of the signal
2 Assign a fixed part of the signal samples to the training one
3 Assign the signal sample to the training one with a certain probability

Answer 1

It is impossible to answer the question without knowing the purpose of the study.
Well, for example - "Relate the signal count to the training one with a certain probability" - whether you can do this or not depends on whether your "signal" depends on time, i.e. is it a time series, or just arbitrary measurements of a certain parameter, for example, obtained from different objects?
"The fixed part of the signal counts is assigned to the training one" - again, for what and what nature of the data you have. And what does "fixed" mean - in the form of a certain number? Or as a percentage of the total? And in general, to train a neural network, you gathered to train it on a single "signal", or you took a lot of these signals (data sets) and want to use some of them for training,
"Take a section of the signal" - is it somehow different from "take a fixed part of the samples"?
In general, without understanding the task, giving you advice is like giving advice to a doctor on how to be treated over the phone - at least unprofessionally, and often it can simply be harmful (in your case, it can make you waste your resources using incorrect methods).