启发式搜索策略应用研究及认知系统仿真  

Strategies application of heuristic search and cognitive system simulation

作　　者：王日凤[1,2] 钟宁[1] 

机构地区：[1]北京工业大学国际WIC研究院,北京100124 [2]广西工学院计算机工程系,广西柳州545006

出　　处：《计算机工程与设计》2011年第12期4172-4177,共6页Computer Engineering and Design

基　　金：国家自然科学基金项目(60875075)

摘　　要：为了寻找人类智能在启发式搜索上的优势,帮助建立基于认知的网络搜索模型,提出了基于视觉感知和目标递归的综合策略。采用功能技术fMRI结合ACT-R认知系统仿真的方法。以简化四方趣题为范式,通过fMRI实验获取信息加工同步的脑激活模式。分析了问题求解过程提出视觉感知策略和目标递归策略混合策略假设,并建立产生式系统。ACT-R认知系统模拟启发式搜索中综合策略应用过程,仿真实验结果表明,反应时偏差0.3s,脑功能区BOLD效应拟合值0.95。研究结果表明,综合策略应用能实现信息有效选择、缩短加工路径、减轻记忆负担和加速搜索过程,是人在启发式搜索上区别于机器的优势。To find out the superiority of human heuristic search to set up new web search model based on cognition,an integrate strategy combined visual perception and goal recursion is proposed.A method of（fMRI functional magnetic resonance imaging） meeting ACT-R（adaptive control of thought-rational） is used.An fMRI experiment on simplified 4€？ Sudoku paradigm is preformed to get the activation pattern of brain regions associated to the mental processing.Based on tasks analysis,an integrate strategy combined visual perception strategy and goal recursion strategy is assumed and a production system is presented.Based on the procedural system,an ACT-R cognitive system is programmed to simulate the integrated strategy used.Compared ACT-R prediction with fMRI result,difference on response time is 0.3 seconds and the average correlation on BOLD（blood oxygen level-dependent） response is about 0.95.It is indicated that integrated strategies is one of superiority of human heuristic search,which help to select efficient information,shorten the scale of processing,weaken the memory burden and speed up search.

关 键 词：问题解决 启发式搜索 策略应用 认知模型 功能磁共振成像 

分 类 号：TP391.9[自动化与计算机技术—计算机应用技术]