知识与数据双向驱动的多粒度认知计算  

A multi-granularity cognitive computing model bidirectionally driven by knowledge and data

作　　者：王国胤[1] 李帅[1] 杨洁[1] WANG Guoyin;LI Shuai;YANG Jie(Chongqing Key Laboratory of Computational Intelligence,Chongqing University of Posts and Telecommunications,Chongqing 400065,Chin)

机构地区：[1]重庆邮电大学计算智能重庆市重点实验室,重庆400065

出　　处：《西北大学学报（自然科学版）》2018年第4期488-500,共13页Journal of Northwest University（Natural Science Edition）

基　　金：国家重点研发计划资助项目(2016QY01W0200);国家自然科学基金资助项目(61772096;61572091)

摘　　要：认知计算是以人类认知为启发而提出的一种一致的、统一的、普遍的智能计算机制。文中介绍一种借鉴人类智能问题求解的多粒度思维机制、人类大脑"大范围优先"的认知机制和智能控制系统中"智能计算前置"的信息处理机制,实现知识与数据双向驱动的多粒度认知计算模型,即数据驱动的粒认知计算模型(DGCC模型),实现符合人类认知的智能计算。在该模型中,数据被视为在多粒度空间中对知识的最细粒度层次表达,知识被视为数据在粗粒度层次的抽象。文中从知识驱动和数据驱动两个方向讨论DGCC模型中需要进一步研究的几个科学问题,并分析几个相关的多粒度认知计算应用案例。Cognitive computing aims to develop a coherent, unified, universal mechanism inspired by human mind′s capabilities. In this paper, the data-driven granular cognitive computing （DGCC）, which is a new cognitive computing model, will be introduced. It is inspired by the multi-granularity thinking mechanism of human intelligence problem solving, the human cognition mechanism of ＂global precedence＂ and the information processing mechanism of ＂intelligent computation forwarding＂ of intelligent control systems. It is a computing mechanism bidirectionally driven by knowledge and data.It tries to implement intelligent computing conforming to human cognition.In DGCC, data is taken as knowledge expressed in the lowest granularity level of a multiple granularity space, while knowledge as the abstraction of data in coarse granularity levels. The scientific research issues of DGCC to be further studied are discussed according to both data-driven and knowledge driven fields. Some application examples are analyzed.

关 键 词：粒认知计算 数据驱动 知识驱动 粒计算 认知计算 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程]