Personalized Re-ranking

In the industry there is a trend to add a re-ranker at the final stage of a recommendation system. The re-ranker ranks the items that have already been filtered out from an enormous candidate set, aiming to provide the finest level of personalized ordering before the items are ultimately delivered to the user. In this …

Practical considerations of off-policy policy gradient

I’d like to talk more about policy gradient [1], which I touched upon in 2017. In common online tutorials, policy gradient theorem takes a lot of spaces to prove that the gradient of the policy in the direction to improve accumulated returns is: where is the accumulated return beginning from step from real samples.  Note …