I guess it's mainly due to patents and such. I'm also pretty sure even the developers don't understand exactly how the algorithms work. sure they modeled some factors into it, probably too many for one person to fully understand on their own...
but once you start crunching immense amount of data it's unavoidable to get unexpected results. and must be impractical to fully monitor.
Well, someone obviously must understand how the algorithms work otherwise they couldn't have implemented it. I actually don't think the algorithms are that complicated and have been thinking about writing a toy simulation to demonstrate how reinforcement learning can be used to re-order the sequence of interactions in a social media sandbox in order to keep users engaged for as long as possible.
I just wanted to know if there was prior work on this that I could use before I went ahead and did the work because if someone has already done this then it would save me a lot of hassle. But it doesn't look like the question will get an answer on cs.stackexchange so I'll probably have to do the work myself.
2 comments
[ 0.18 ms ] story [ 17.7 ms ] threadI just wanted to know if there was prior work on this that I could use before I went ahead and did the work because if someone has already done this then it would save me a lot of hassle. But it doesn't look like the question will get an answer on cs.stackexchange so I'll probably have to do the work myself.