I saw this comment from YouTube user mccallr, and I thought this would be a good discussion to have.
would be helpful to hear about the pros and cons of the scalar encoder vs. RDSE. I’m wondering in what situation I’d want to use one over the other.
So I’ll pose this question to you all. In what situations would you use the normal scalar encoder vs the Random Distributed Scalar Encoder? What about @mrcslws’s sampling linear encoder? How about a delta encoder (we haven’t talked about that)?
Are they any advantages in using a RDSE?
I see downsides in needing to calculate representations for contiguous values which have not been encountered in order to represent a new value.
The main advantage is that you don’t need to bound the input. With a scalar encoder, you must provide min/max values. But with RDSE you can fudge it. See this video for visuals. (sorry about the hair it was a phase)