Toshinori Kitamura, Tadashi Kozuno, Yunhao Tang, Nino Vieillard, Michal Valko, Wenhao Yang, Jincheng Mei, Pierre Ménard, Mohammad Gheshlaghi Azar, Remi Munos, Olivier Pietquin, Matthieu Geist, Csaba Szepesvari, Wataru Kumagai, Yutaka Matsuo. “Regularization and Variance-Weighted Regression Achieves Minimax Optimality in Linear MDPs: Theory and Practice”, International Conference on Machine Learning (ICML 2023)
Mirror descent value iteration (MDVI), an abstraction of Kullback–Leibler (KL) and entropy-regularized reinforcement learning (RL), has served as the basis for recent high-performing practical RL algorithms. However, despite the use of function approximation in practice, the theoretical understanding of MDVI has been limited to tabular Markov decision processes (MDPs). We study MDVI with linear function approximation through its sample complexity required to identify an $\varepsilon$-optimal policy with probability $1-\delta$ under the settings of an infinite-horizon linear MDP, generative model, and G-optimal design. We demonstrate that least-squares regression weighted by the variance of an estimated optimal value function of the next state is crucial to achieving minimax optimality. Based on this observation, we present Variance-Weighted Least-Squares MDVI (VWLS-MDVI), the first theoretical algorithm that achieves nearly minimax optimal sample complexity for infinite-horizon linear MDPs. Furthermore, we propose a practical VWLS algorithm for value-based deep RL, Deep Variance Weighting (DVW). Our experiments demonstrate that DVW improves the performance of popular value-based deep RL algorithms on a set of MinAtar benchmarks.
Shohei Taniguchi, Masahiro Suzuki, Yusuke Iwasawa, Yutaka Matsuo. “End-to-end Training of Deep Boltzmann Machines by Unbiased Contrastive Divergence with Local Mode Initialization”, International Conference on Machine Learning (ICML 2023)
We address the problem of biased gradient estimation in deep Boltzmann machines (DBMs). The existing method to obtain an unbiased estimator uses a maximal coupling based on a Gibbs sampler, but when the state is high-dimensional, it takes a long time to converge. In this study, we propose to use a coupling based on the Metropolis-Hastings (MH) and to initialize the state around a local mode of the target distribution. Because of the propensity of MH to reject proposals, the coupling tends to converge in only one step with a high probability, leading to high efficiency. We find that our method allows DBMs to be trained in an end-to-end fashion without greedy pretraining. We also propose some practical techniques to further improve the performance of DBMs. We empirically demonstrate that our training algorithm enables DBMs to show comparable generative performance to other deep generative models, achieving the FID score of 10.33 for MNIST.