In this talk, we will consider the problem of binary hypothesis testing with finite memory. Consider a sequence of IID random variables with expectation p under hypothesis H_0 and q under hypothesis H_1.

Multi-armed bandit (MAB) is a popular framework for decision-making in an uncertain environment. In its classical setup, the algorithm has access to a fixed and finite set of unknown, independent probability distributions or arms.

We will try to understand the paper "Bipartite Perfect Matching is in quasi-NC" by Fenner, Gurjar and Thierauf (2016).