Industrial Benchmark 弥合人工实际问题属性数据集

Industial Benchmark 是一个弥合免费的、记录的和主动的人工问题与实际工业问题属性之间的差距的数据集。数据集有关线下 RL 和线上 RL 的现实，用于查找最适合现实世界应用程序的 RL 算法。数据集旨在提供可解释的 RL 培训场景，并详细了解方法的学习过程。（通过在 Github 上发布其 Java 和 Python 代码，包括 OpenAI Gym 包装器，产生的工业数据（IB）已公开提供给 RL 社区）

Industrial Benchmark is a dataset that bridges the gap between free, documented, and proactive artificial benchmark problems and the attributes of real-world industrial problems. The dataset covers real-world scenarios for both offline reinforcement learning (RL) and online RL, aiming to identify the most suitable RL algorithms for real-world applications. This dataset is designed to provide interpretable RL training scenarios and enable in-depth insights into the learning process of the methods. The resulting industrial benchmark (IB) data has been publicly released to the RL community by publishing its Java and Python code (including OpenAI Gym wrappers) on GitHub.

搜集汇总

背景与挑战

背景概述

Industrial Benchmark 是一个旨在弥合人工问题与实际工业问题属性之间差距的数据集，专注于线下和线上强化学习（RL）的现实应用，以帮助找到最适合现实世界场景的RL算法。该数据集提供可解释的RL训练场景，并详细描述方法的学习过程，其代码已在Github上公开。