Multi-armed bandit strategies are primarily used in the areas of artificial intelligence, big data, smart data and digital marketing. They help to make decisions when many possibilities need to be tested simultaneously.
Imagine a slot machine with several levers ("arms"). Each lever pays out a profit at a different frequency, but you don't know which one is best at the beginning. This is what multi-armed bandit strategies are all about: as you try out which lever gives the highest return, you learn more and more which one is the best choice. Instead of testing all options equally often, you gradually concentrate on the most promising options.
A practical example: In digital marketing, a website can display two different advertising banners. With a multi-armed bandit strategy, both banners are initially displayed equally often. If the system recognises that one of the banners is particularly popular, it is displayed more frequently. This constantly optimises the advertising effect and saves your company time and advertising budget.
To summarise: With multi-armed bandit strategies, companies can use data to find out more quickly which alternative works best - without unnecessary experimentation and inefficiency.
Deutsch 
English (UK)