Randomized Exploration for Reinforcement Learning with Multinomial Logistic Function Approximation

Reinforcement learning (RL) is a sequential decision-making problem in which an agent tries to maximize its expected cumulative reward by interacting with an unknown environment over time.