[fea] Simplify our beta judgement rule

[yiakwy]

Hi, I have scanned through the paper and already dive into the implementation part to observe data processing flow.

I found that our beta judgment rule

could be simplified into the following format with less numeric redundancies:

Because theta should not be large and positive, this method is much better looking for your paper.

[niosus]

Thanks for your suggestion. Can you elaborate a bit more where you got this formula from? I don't see the reasoning behind it.

[niosus]

Closing for the lack of activity.