Case-Based Goal Formulation

Case-Based Goal Formulation
复制标题

基于案例的目标制定

DOI:
--
复制
发表时间:
2010
期刊:
影响因子:
--
通讯作者:
A. Jhala
A. Jhala
中科院分区:
--
文献类型:
--
作者:
B. Weber;Michael Mateas;A. Jhala

文献摘要

被引文献

相似文献

强大的人工智能系统需要能够推理它们的目标,并根据给定的情况制定新的目标。基于案例的目标制定是一种使用示例库为代理制定新目标的技术。我们提供了该术语的形式化以及实现该定义的两个算法。在实时战略游戏《星际争霸》的对手建模和策略选择任务上,将该算法与基于实例和基于模型的技术进行了比较。我们的系统EISBot实现了这些技术,并且能够持续击败星际争霸的内置AI。
Robust AI systems need to be able to reason about their goals and formulate new goals based on the given situation. Case-based goal formulation is a technique for formulating new goals for an agent using a library of examples. We provide a formalization of this term and two algorithms that implement this definition. The algorithms are compared against instance-based and modelbased techniques on the tasks of opponent modeling and strategy selection in the real-time strategy game StarCraft. Our system, EISBot, implements these techniques and is capable of consistently defeating the builtin AI of StarCraft.