arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2401.04290cs.CVcs.AIcs.MA
收藏

StarCraftImage:用于多智能体环境空间推理方法原型设计的多智能体环境数据集

StarCraftImage: A Dataset For Prototyping Spatial Reasoning Methods For Multi-Agent Environments

Purdue University(普渡大学);DEVCOM Army Research Laboratory(DEVCOM陆军研究实验室);ARL(陆军研究实验室)

Sean Kulinski, Nicholas R. Waytowich, James Z. Hare, David I. Inouye

AI总结:

本文基于StarCraft II回放构建了易于使用的基准空间推理数据集StarCraftImage,从6万场回放中提取360万张摘要图像并提供三种复杂度递减的格式,以支持多智能体环境空间推理方法的快速原型设计与复现。

摘要:

多智能体环境中的空间推理任务(如事件预测、智能体类型识别或缺失数据填补)对多种应用(例如传感器网络上的自主监控,以及强化学习(RL)的子任务)非常重要。StarCraft II 游戏回放编码了智能(及对抗性)的多智能体行为,可为这些任务提供测试平台;然而,提取简单且标准化的表示以进行这些任务的原型设计非常费力,且阻碍了可复现性。相比之下,MNIST 和 CIFAR10 尽管极其简单,却使 ML 方法的快速原型设计和复现成为可能。遵循这些数据集的简单性,我们基于 StarCraft II 回放构建了一个基准空间推理数据集,该数据集展现了复杂的多智能体行为,同时仍像 MNIST 和 CIFAR10 一样易于使用。具体而言,我们仔细总结了 255 个连续游戏状态的窗口,从 60,000 场回放中创建了 360 万张摘要图像,包括所有相关的元数据,如游戏结果和玩家种族。我们开发了三种复杂度递减的格式:包含每种单位类型一个通道的高光谱图像(类似于多光谱地理空间图像)、模仿 CIFAR10 的 RGB 图像,以及模仿 MNIST 的灰度图像。我们展示了如何使用该数据集进行空间推理方法的原型设计。所有数据集、提取代码和数据集加载代码可在 https://starcraftdata.davidinouye.com 找到。

英文摘要:

Spatial reasoning tasks in multi-agent environments such as event prediction, agent type identification, or missing data imputation are important for multiple applications (e.g., autonomous surveillance over sensor networks and subtasks for reinforcement learning (RL)). StarCraft II game replays encode intelligent (and adversarial) multi-agent behavior and could provide a testbed for these tasks; however, extracting simple and standardized representations for prototyping these tasks is laborious and hinders reproducibility. In contrast, MNIST and CIFAR10, despite their extreme simplicity, have enabled rapid prototyping and reproducibility of ML methods. Following the simplicity of these datasets, we construct a benchmark spatial reasoning dataset based on StarCraft II replays that exhibit complex multi-agent behaviors, while still being as easy to use as MNIST and CIFAR10. Specifically, we carefully summarize a window of 255 consecutive game states to create 3.6 million summary images from 60,000 replays, including all relevant metadata such as game outcome and player races. We develop three formats of decreasing complexity: Hyperspectral images that include one channel for every unit type (similar to multispectral geospatial images), RGB images that mimic CIFAR10, and grayscale images that mimic MNIST. We show how this dataset can be used for prototyping spatial reasoning methods. All datasets, code for extraction, and code for dataset loading can be found at https://starcraftdata.davidinouye.com

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑