RI: Small: Cooperative Coevolutionary Design and Multiagent Systems

协同进化是解决大规模随机优化问题的有效途径。这种计算方法中固有的未解决的博弈论挑战是复杂的，对进化计算社区具有重大意义。该项目正在推进共同进化的最新技术，并将其应用于比文献中常见的问题大得多的问题。这些挑战及其解决方案对其他协同适应环境（如多智能体强化学习、分布估计算法、智能体建模和群机器人）具有潜在的变革性影响。协同进化在使用多智能体系统模型的领域有很强的适用性，包括多机器人、生物学、经济学、土地利用和政治学。这些领域的更好模式可以对社会、政策、国土安全和环境产生积极影响。

Cooperative coevolution is a potent approach to doing large-scale stochastic optimization. The unsolved game-theoretic challenges inherent in this computational method are complex and of significant interest to the evolutionary computation community. This project is advancing the state of the art in coevolution and is applying it to significantly larger problems than commonly found in the literature. These challenges, and their solution, have potentially transformative impact on other co-adaptive environments such as multiagent reinforcement learning, estimation of distribution algorithms, agent modeling, and swarm robotics. Coevolution has strong applicability to fields that use multiagent system models, including multirobotics, biology, economics, land use, and political science. Better models in these fields can positively affect society, policy, homeland security, and the environment.