arXivDaily arXiv每日学术速递 周一至周五更新
北美退伍军人节(Veterans Day),12 日无更新
arXiv 1706.01458physics.soc-phcond-mat.softcond-mat.stat-mech
收藏

智能体的智能在集体行为中产生结构性的相变

Intelligence of agents produces a structural phase transition in collective behaviour

Hannes Hornischer, Stephan Herminghaus, Marco G. Mazza

摘要:

生物体处理信息以相互作用并适应不断变化的环境,目的是寻找食物,交配或避免危险。它们的生态位结构通过选择它们的内部结构和诱导对环境线索和刺激的适应性反应而产生深远的影响。由专门的优化策略支持的适应性集体行为在自然界中无处不在。这种非凡的成功源于适应和选择的过程。在这里,我们证明了一个普遍的物理机制的非平衡过渡的信息处理有机体的集体组织的基础。当认知主体建立并更新其环境的因果结构的内部认知表征时,复杂模式在系统中出现,其中斑图形成的开始与认知地图的空间重叠有关。研究代理之间的信息交换揭示了一个连续的,有序无序的过渡。由于平移对称性的自发破缺,出现了戈德斯通模式,它指出了认知有机体之间信息传递的集体机制。总而言之,这种相变的特征以普遍的方式巩固了认知科学和生物科学中的不同结果。这些发现通常适用于不依赖于集中控制方案的人工智能群体系统的设计。

英文摘要:

Living organisms process information to interact and adapt to their changing environment with the goal of finding food, mates or averting hazards. The structure of their niche has profound repercussions by both selecting their internal architecture and also inducing adaptive responses to environmental cues and stimuli. Adaptive, collective behaviour underpinned by specialized optimization strategies is ubiquitously found in the natural world. This exceptional success originates from the processes of fitness and selection. Here we prove that a universal physical mechanism of a nonequilibrium transition underlies the collective organization of information-processing organisms. As cognitive agents build and update an internal, cognitive representation of the causal structure of their environment, complex patterns emerge in the system, where the onset of pattern formation relates to the spatial overlap of cognitive maps. Studying the exchange of information among the agents reveals a continuous, order-disorder transition. As a result of the spontaneous breaking of translational symmetry, a Goldstone mode emerges, which points at a collective mechanism of information transfer among cognitive organisms. Taken together, the characteristics of this phase transition consolidate different results in cognitive and biological sciences in a universal manner. These finding are generally applicable to the design of artificial intelligent swarm systems that do not rely on centralized control schemes.

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。

↑