WE can use UCB, Thompson sampling for MAB problems.
Similarly, for contextual bandits, we can use LinUCB and LinTS.
However, why doesnot the vowpal wabbit library support LinUCB, LinTS for exploration algorithms ??
IS it true that UCB, TS based approaches doesn't work if we have rich context as mentioned here?? (""....UCB and TS based approaches are only good when you have relatively little contextual information available....."" )
https://gitter.im/VowpalWabbit/community?at=5e5c2c2eec379e558e9c627b
I saw lot of papers and literature where thompson sampling is being applied to contextual bandits, though.
