GridWorld Problem

该数据集包含了多种GridWorld问题的配置，用于评估所提出的方法。每种配置在动作数量、状态、障碍物以及终止状态等方面都有所不同。该数据集还用于比较CP-MDP方法与价值迭代和策略迭代算法的标准表格实现的效率。数据集包含了多达一百万个状态以及不同维度的配置。其任务是通过张量分解方法解决马尔可夫决策过程问题。

This dataset contains configurations for various GridWorld problems, which are used to evaluate the proposed method. Each configuration differs in terms of the number of actions, states, obstacles, terminal states, and other relevant aspects. Additionally, this dataset is employed to compare the efficiency of the CP-MDP method with the standard tabular implementations of value iteration and policy iteration algorithms. The dataset includes up to one million states and configurations with varying dimensions. The core task of this dataset is to solve Markov Decision Process (MDP) problems via tensor decomposition methods.