The technique might look a bit annoying at first glance, but it clearly minimizes the error rate for users who delete sensitive data by mistake.

The network model can organize and study itself, according to different study error, continuously adjust the study rate, and accelerate refrain process, expel influence of the data sample.

But even after I sorted through the data to make the typed responses consistent with the menu responses, there was still a difference in error rate.