Yoel Zeldes • 8/14/2018

Recommender Systems: Exploring the Unknown Using Uncertainty

This technical article discusses the exploration-exploitation trade-off in recommender systems, using the example of Click-Through Rate (CTR) prediction. It explains basic (ε-greedy) and advanced (Upper Confidence Bound) algorithms that use uncertainty to intelligently recommend new content while maximizing user clicks.

0 comments

#Reinforcement Learning #Click Through Rate #Recommender Systems