Linear bandits is a research area within machine learning and reinforcement learning that focuses on the problem of sequentially making decisions in a multi-armed bandit setting where the reward for choosing an action is linear in nature. In other words, the rewards of different actions are assumed to be a linear function of some unknown parameters that need to be estimated by the decision-making algorithm. The goal of linear bandit algorithms is to balance the exploration of different actions to learn these parameters with the exploitation of actions that are likely to yield high rewards based on the current estimates. Linear bandits find applications in a variety of domains, including online advertising, recommendation systems, and personalized content delivery.