Overview of EE (Exploitation-Exploration) Problem
The EE (Exploitation & Exploration) problem is very common in computational advertising/recommender systems. More broadly, any decision-making problem involves the EE problem. Simply put, this problem addresses whether to choose the optimal strategy based on existing experience (Exploitation) or explore new strategies to improve future returns (Exploration) when making decisions. This article mainly introduces three common methods to solve this problem: random methods, UCB methods, and Thompson sampling methods, focusing on the specific procedures and basic ideas of each method.