I am learning about Reinforcement learning in the book Grokking Deep Reinforcement Learning. Below are snippets. Below is the description of Bandit Slippery Walk (BSW)

enter image description here

Below is the description of two arm Bernoulli Bandit Environment

enter image description here

I am having difficulty in understanding the following paragraph

This is I i.e., Bernoulli Bandit Environment) similar to the BSW to an extent. BSW has complimentary probabilities: action 0 pays +1 with α probability, and action 1 pays +1 with 1–α chance. In this kind of bandit environment, these probabilities are independent; they can even be equal.

Why does it state that BSW has complimentary probabilities? If action 0 has $\alpha$ probability for +1 reward, why does action 1 pays +1 with $1- \alpha$, as action 1 is independent of action 0?


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.