Randomized Exploration for Reinforcement Learning with Multinomial Logistic Function Approximation

TL;DR

This paper introduces efficient randomized algorithms for reinforcement learning with multinomial logistic transition models, providing theoretical regret guarantees and demonstrating superior empirical performance.

Contribution

The paper develops the first randomized RL algorithms with constant-time per-episode computation for MNL transition models, with provable regret bounds.

Findings

RRL-MNL achieves a regret of old rac{1}{\u03ba} d^{3/2} H^{3/2} sqrt{T}

ORRL-MNL improves regret dependence on rac{1}{ba} with an additional term

Numerical experiments show the algorithms outperform existing methods in practice

Abstract

We study reinforcement learning with multinomial logistic (MNL) function approximation where the underlying transition probability kernel of the Markov decision processes (MDPs) is parametrized by an unknown transition core with features of state and action. For the finite horizon episodic setting with inhomogeneous state transitions, we propose provably efficient algorithms with randomized exploration having frequentist regret guarantees. For our first algorithm, $\texttt{RRL-MNL}$, we adapt optimistic sampling to ensure the optimism of the estimated value function with sufficient frequency. We establish that $\texttt{RRL-MNL}$ achieves a $\tilde{O}(\kappa^{-1} d^{\frac{3}{2}} H^{\frac{3}{2}} \sqrt{T})$ frequentist regret bound with constant-time computational cost per episode. Here, $d$ is the dimension of the transition core, $H$ is the horizon length, $T$ is the total number of steps, and $\kappa$ is a problem-dependent constant. Despite the simplicity and practicality of $\texttt{RRL-MNL}$, its regret bound scales with $\kappa^{-1}$, which is potentially large in the worst case. To improve the dependence on $\kappa^{-1}$, we propose $\texttt{ORRL-MNL}$, which estimates the value function using the local gradient information of the MNL transition model. We show that its frequentist regret bound is $\tilde{O}(d^{\frac{3}{2}} H^{\frac{3}{2}} \sqrt{T} + \kappa^{-1} d^2 H^2)$. To the best of our knowledge, these are the first randomized RL algorithms for the MNL transition model that achieve statistical guarantees with constant-time computational cost per episode. Numerical experiments demonstrate the superior performance of the proposed algorithms.