It’s a SRS without context. You could probably put contextual bandits somewhere in the loop and optimize for the individual. I found it very disappointing when

It’s a SRS without context. You could probably put contextual bandits somewhere in the loop and optimize for the individual. I found it very disappointing when Duolingo released their paper on optimizing their global SRS on data they collected from the users. There was no context, only curve fitting for a better multipler of fail and success.