关键词： 机器学习   条件概率   层次分类   域   人工智能  

摘要： 本文通过对著名站点Yahoo层次结构的分析，介绍一种基于此层次结构的文档分类器的机器学习技巧。我们利用此层次学习算法来自动构造分类器每个域的背景知识。

关键词： 人工智能   机器学习   智能体   协调   市场   竞争  

摘要： 该文目的是构造一种通用机器学习系统,这种系统集成了多种机器学习方法,适合不同的学习任务,能够适应复杂、大型环境的学习.具体做法是:在系统中设立市场调节机构,资金和知识通过市场进行交换;各学习机采用函数形式表示其活动规律,行动具有主动性并遵循一定的行为规则,目标是资金积累最大化.各学习机购买系统资源进行学习,按照学习得到的知识质量获得报酬,然后进行下一阶段的学习,如此不断循环.在循环过程中,适合的学习机资金积累逐渐增加,活动能力增强,不适合的学习机资金积累逐渐减少,活动性减弱,这样通过优胜劣汰只保留适合的学习方法,系统资源得到了合理分配,并且系统可以达到通用、自适应、自优化的要求.文章最后分析了系统的稳定性.

关键词： 机器翻译   机器学习   搭配规则  

摘要： 本文针对汉英机译系统(CETRAN)中搭配词典的构造,讨论了词语搭配的特点、形式描述以及搭配规则的自动获取问题,提出了一种由词语搭配关系获取机器翻译系统中词类搭配规则的机器学习方法(CRLM)。该方法在对大规模语料进行统计的基础上,找出词语间词类的关联例并引入机器学习技术,构造词类搭配树,获取用于词类有序消歧的搭配规则。

关键词： Crystallization  

摘要： A revised machine learning methodology is applied to a simulated crystallization process flowsheet for continual improvement of its performance by generating and analyzing process data. The aim is to identify bands of crucial decision variables leading to zones of best average process performance. The methodology comprises two components: symbolic induction and case-based reasoning. It uses an incremental algorithm to update performance classification rules, which not only improves the efficiency of the symbolic induction of the classification rules by eliminating the need for their periodic reinduction but also simplifies the case-based reasoning step. The new concepts and procedures are illustrated by application to a potassium nitrate crystallization process comprising a mixed-suspension, mixed-product-removal crystallizer, a hydrocyclone, and a fines dissolver. By identifying and establishing ranges for the most crucial decision variables, namely feed concentration, flow rate, and cooling stream temperature, three zones leaning to an improvement of nearly 12% on nominal average performance are detected within two generations of the classification rules.

关键词： 分类器系统   机器学习   环境标记   遗传算法  

摘要： 该文首先介绍了分类器系统的基本原理,然后对它的一个具体实现-零级分类器系统(ZCS)及其试验环境进行了细致的描述;在此基础上,该文主要完成了三项研究内容:1)针对ZCS结构上的不足进行了相应的改进,使得它在Woods环境下的学习达到了最优值;2)在自然界中存在的\"主动学习\"现象的启发下,抽象出\"环境标记\"的概念,并提出\"基于环境标记的分类器系统\"框架,进而用此框架模型对ZCS进行了扩展;扩展后的ZCS在\"动态Woods\"环境下与其它系统的比较试验结果表明,引入环境标记处理能力对提高分类器系统的学习能力是切实有效的;3)尝试用基于环境标记的分类器系统实现了一个简单的多智能机协作系统.

关键词： DECISION support systems   ARTIFICIAL intelligence   MACHINE theory   MANAGEMENT information systems   MACHINE learning   TELEMATICS  

摘要： When a decision maker has access to multiple expert systems, each embodying a different expert perspective on analyzing and reasoning about the same bind of decision problem, an important consideration is which to use at what times. We address this issue with a method based on competition among the distinct expert systems (and their respective rules). We begin by reviewing prior research concerned with the coordination of multiple sources of expertise in support of decision making, pointing out potential weaknesses of the proposed methods. Next, we introduce a new coordination method based on the competitive paradigm that has been applied in machine learning. This method involves adjustments to the strengths of expert systems and to their constituent rules based on their performances. A nine-step process for adjusting strengths is described. Advantages and limitations of this new method for expert system coordination are discussed. We outline an approach to testing the coordination method and report on preliminary testing of the performance of a system employing our method versus the performance of individual experts.

关键词： 命题逻辑   有向无环图   自动定理证明   可满足性   机器学习   算法复杂性   谓词   归纳   代数递归  

摘要： 该文就人工智能中两个重要的领域——命题逻辑的自动定理证明与机器学习——进行了一定程度的探讨.该文提出了基于一类特殊的有向无环图OSTD的命题公式表示、判断命题公式可满足性和通过真值表学习命题公式的方法.在自动定理证明领域,命题公式的可满足性是一个典型的NP—完全问题.该文提出:以OSTD作为命题公式的内部表示形式,采用图的路径搜索来完成定理的证明,公式的可满足性判断,和从真值表中学习命题公式.并证明,上述工作可以在多项式时间内完成.同时,在PC机Windows平台下,用Visual Basic初步实现了该文中提出的方法.进一步地,该文讨论了带有参数的命题,即谓词知识的机器学习.并且具体研究了代数递归谓词的归纳学习.最后指出了今后进一步需要研究的课题.

关键词： Human-computer interaction   Machine learning  

摘要： Intelligent assistant systems provide an adequate organization of human-computer interaction for complex problem solving. These knowledge-based systems are characterized by a cooperative problem-solving procedure. User and system cooperate intensively to produce the aimed result. Machine learning methods can provide significant support for assistant systems. In this article, it is pointed out how assistant systems can be supported in various ways. For instance, machine learning methods can extend, revise, optimize, and adapt the knowledge base of an assistant system. In this way, they can contribute to the utility and maintainability of an intelligent assistant system. They can also increase the flexibility and effectiveness of human-computer interaction. The learning apprentice system COSIMA is presented which acquires knowledge about single problem-solving steps from observation of the user. Production rules for floorplanning, a sub-task of VLSI design, are acquired and refined cooperatively by different learning strategies.

关键词： 机器学习   知识库   人工智能   归纳机器学习   RVPA  

摘要： 为了支持在事实不完全或不充分环境中的有效推理，作者提出了一种归纳机器学习方法，并设计了一个规则向量投影算法，使用木文介绍的算法可对原始知识实行归纳，生成含一系列全新分类概念和推理路经的网络知识库，基于该知识库的机器推理系统，在作出诊断决策时所需事实量可大为减少，因此在信息量不足的情况下仍能具有很高的推理性能．

关键词： 学习率   学习效率   CAD   土木工程  

摘要： 如果CAD系统具有智能,就可以提高设计的质量和效率,缩短设计周期,成为设计者的顾问和助手,但是,专家知识获取困难.解决知识获取的有效途径是建立学习系统.其次知识需要不断地积累.没有学习功能的智能系统,它只是建造时输入的知识,不会发展和创新,面对新情况,将会束手无策.因此,一个系统没有学习知识的能力,就不具备真正的智能.