开发、维护和检验知识库的困难激励我们对自适应系统更感兴趣。但是,已经证明符号范例有碍于动态或噪声环境中的学习。本报告叙述基于仿真的研究计划,该计划说明作为在这样的环境中学习的健壮增强模型基础的修正连接式记忆单元的效用。这种学习模型根据一个定性的、基于目标的结构产生有效的基于规则的行为,这种结构进一步支援从这些复杂领域的经验中导出的二级目标或智力模型的开发。 The difficulty of developing, maintaining and verifying the knowledge base motivated us to be more interested in adaptive systems. However, the paradigm of symbols has proven to be a deterrent to learning in dynamic or noisy environments. This report describes a simulation-based research project that illustrates the utility of modifying connected-connected memory cells as the basis for robust augmentation models learned in such environments. This learning model generates effective rules-based behavior based on a qualitative, goal-based structure that further supports the development of secondary goals or mental models derived from the experience of these complex areas.