群智演化协同计算的研究进展与趋势

Advances and Trends in Crowd Intelligence Evolutionary and Collaborative Computation

  • 摘要: 群体智能是通过聚集群体智慧协同求解大规模复杂问题的智能方法,其思想最初源于对自然界中社会性生物群体智能行为的模拟。群体生物通过分工合作、相互协调、协同演化,可涌现出整体性的智能行为,完成复杂任务,具有高度的自组织、自适应、自学习能力。受此启发,国内外学者运用数学和计算机工具对群体智能行为进行模拟,从不同角度发展了一系列群体智能涌现与演化的机理和模型。近年来,随着互联网的发展,人类社会基于物联网的群智协同和演化现象进一步拓宽了群智演化计算的范畴,呈现出广阔的应用前景,也对群智演化的理论模型和应用提出了新挑战。2017年,《新一代人工智能发展规划》明确将群体智能列为需重点发展的人工智能理论与技术方向之一。文章将从生物群体、智能体群体和人类社会群体等不同视角,从群智演化协作的模型和机理、群智演化协作的组织结构、群智演化协同决策及群智演化协同计算的应用等角度,总结群智演化计算的主要研究问题,对国内外的最新研究进展进行综述和对比分析,并对该方向未来的发展趋势和主要科学问题进行展望。

     

    Abstract: Crowd intelligence is an intelligent method for solving large-scale complex problems by aggregating group intelligence. Its idea is originated from the simulation of the intelligent behaviors of social biological swarms in nature. Through the division of labor, cooperation, coordination, and co-evolution, swarm creatures can emerge integrated intelligent behaviors. Swarms can complete complex tasks with a high degree of self-organization, self-ada-ptation and self-learning ability. Inspired by this, researchers use mathematics and computing tools to simulate the behavior of swarm intelligence, and develop a series of mechanisms and models based on the emergence and evolution of swarm intelligence. In recent years, with the development of the Internet, the collaborative and evolutionary phenomenon of human crowd intelligence based on the Internet of Things has further broadened the scope of crowd intelligence, presenting a broad application prospect, which also poses new challenges to the theoretical models and applications of group intelligence evolution. In 2017, "New Generation Artificial Intelligence Development Plan" clearly listed crowd intelligence as one of the important artificial intelligence theories and techniques to be deve-loped. In this report, the development of crowd intelligence has been studied from different perspectives: from biological swarms to multi-agent systems, and further to human crowds. The main research issues in crowd intelligence and evolutionary computing will be discussed in four aspects: theories and models of crowd intelligence, organization of crowds, crowd intelligence for collaborative decision-making, and the applications of crowd intelligence and evolutionary computing. The main research issues of crowd intelligence and evolutionary computation is summarized in this report. The reviews have also been made for the latest studies in China and abroad. Finally, some discussion has also been made for future trends and potential scientific issues of this research field.

     

/

返回文章
返回