归纳逻辑程序设计中的优化问题研究

归纳逻辑程序设计的核心问题是如何从背景知识中优选谓词构造满足约束的归纳假设,按Ｏｃｃａｍ准则,满足约束的最精简归纳假设为优,但迄今归纳逻辑程序设计中精简归纳假设构造的计算复杂性尚未解决。     

一类递归逻辑程序的强构造学习

定义了一类强构造学习问题ＥＲＰ;探讨了以数据驱动方式提高强构造学习系统效率的途径,特别是学习系统的主动采样方式与系统性能的关系;给出了一个效率较高、输入条件基本实用的ＥＲＰ（１）学习算法,该算法原则上可推广到任意ｎ〉１的ＥＲＰ（ｎ）问题。．     

递归逻辑程序的强构造分层学习算法及其实现

本文提出了一种逻辑程序的强构造分层学习算法用该算法解决了一类递归逻辑程序的强构造二层学问题，为强的构造学习研究提供了一个新的思路，并就有关问题进行了详细的讨论。     

基于归纳的机器学习方法的实现研究

一、引言机器学习是解决计算机获取知识问题的方法。在这项研究中,对人类学习方法的模拟是一条重要的途径,而归纳正是人类学习的重要方式之一。从归纳中获取知识是人类新知识的极其重要的来源,它通过对经验材料的研究,探求新知识。由于一切科学理论的基本前提最初都是从经验事实中总结出来的,而从经验到一般规律的过程中,归纳推理起着不可替代的作用。开普勒从各大行星的天文观测     

构造性归纳综述

1.引言机器学习是人工智能领域的一个长期的研究热点,其研究目标,就是利用计算机来模拟、实现各种形式的学习行为。学习涉及两方面的内容:知识获取和技能获取。前者是从外界获得新的信息,后者是完善已得到的知识。对于这两方面在机器学习领域中有两种主要的学习典范。一种称为符号经验式学习——SEL(Symbolic Empirical Learning)——研究知识获取的本质问题,这种方法试图从事先提供的概念例子中分析出概念相应的规则,或称为描述。这是一种从具体到一般的推理过程,我们称之为归纳,因     

分布式并行约束归纳逻辑程序设计研究

CILP是关系数据挖掘的主要技术之一。为提高CILP系统的效率,提出了一种基于C3模型,元学习技术和主从式静态负载平衡策略的分布式并行CILP算法,并实现了一个基于COW机群结构的分布式并行CILP原型系统。实验表明该算法是高效的,能获得较好的负载平衡,较高的加速比和并行效率。     

Prolog逻辑程序的测试方法

Prolog逻辑程序设计语言具有不同于传统程序设计语言的特点.本文根据测试Prolog程序的实践,提出了一系列为Prolog程序生成测试用例的准则.按这些准则测试实际程序效果良好.这是在逻辑程序测试新领域迈出的一步.     

修正的逻辑程序知识库更新方法

基于逻辑程序的知识库更新方法研究的焦点在于处理知识库的冲突问题,但代价是更新时规则库增大很快,该文提出了“修正的逻辑程序知识库更新方法”,此种方法基于一种规范知识库更新的形式化方法—修正程序,此种更新方法不仅可以最大程度地减少更新时规则库的增大,也避免了重复工作和知识库信息的丢失,还可以同时满足“替换更新”和“丰富更新”。     

逻辑程序的一种并行执行模型

本文基于静态相关性分析和动态调整相结合的方法，提出了一种逻辑程序的执行模型，它不仅开发了“与“并行，同进也开发了一定的“或“并行，从而有效地加速了逻辑程序的执行。     

复杂结构归纳学习研究

复杂结构归纳学习的需求近年来快速增长。复杂结构归纳学习方法按照知识表示方式不同分为基于逻辑的方法与基于数学图的方法。阐述了复杂结构归纳学习研究的历史沿革,介绍、分析和对比了不同知识表示方式下的学习方法,给出了复杂结构归纳学习将来发展面临的挑战和需重点解决的问题。     

Improving the efficiency of inductive logic programming systems

Inductive logic programming (ILP) is a sub‐field of machine learning that provides an excellent framework for multi‐relational data mining applications. The advantages of ILP have been successfully demonstrated in complex and relevant industrial and scientific problems. However, to produce valuable models, ILP systems often require long running times and large amounts of memory. In this paper we address fundamental issues that have direct impact on the efficiency of ILP systems. Namely, we discuss how improvements in the indexing mechanisms of an underlying logic programming system benefit ILP performance. Furthermore, we propose novel data structures to reduce memory requirements and we suggest a new lazy evaluation technique to search the hypothesis space more efficiently. These proposals have been implemented in the April ILP system and evaluated using several well‐known data sets. The results observed show significant improvements in running time without compromising the accuracy of the models generated. Indeed, the combined techniques achieve several order of magnitudes speedup in some data sets. Moreover, memory requirements are reduced in nearly half of the data sets. Copyright © 2008 John Wiley & Sons, Ltd.     

Machine Learning and the foundations of inductive inference

The problem of valid induction could be stated as follows: are we justified in accepting a given hypothesis on the basis of observations that frequently confirm it? The present paper argues that this question is relevant for the understanding of Machine Learning, but insufficient. Recent research in inductive reasoning has prompted another, more fundamental question: there is not just one given rule to be tested, there are a large number of possible rules, and many of these are somehow confirmed by the data — how are we to restrict the space of inductive hypotheses and choose effectively some rules that will probably perform well on future examples? We analyze if and how this problem is approached in standard accounts of induction and show the difficulties that are present. Finally, we suggest that the explanation-based learning approach and related methods of knowledge intensive induction could be, if not a solution, at least a tool for solving some of these problems.     

The strong semantics for logic programs

Recently, the well-founded semantics of a logic programP has been strengthened to the well-founded semantics-by-case (WF_C) and this in turn has been strengthened to the extended well-founded semantics (WF_E). Both WF_C(P) and WF_E(P) have thelogical consequence property, namely, if an atomAj is true in the theory Th(P), thenAj is true in the semantics as well. However, neither WF_C nor WF_E has the GCWA property, i.e., if an atomAj is false in all minimal models ofP,Aj may not be false in WF_C(P) (resp. WF_E(P)). We extend the ideas in WF_C and WF_E to define a strong well-founded semantics WF_S which has the GCWA property. The strong semantics WF_S(P) is defined by combining GCWA with the notion ofderived rules. Here we use a new Type-III derived rules in addition to those used in WF_C and WF_E. The relationship between WF_S and WF_C is also clarified.     

简单单调约束逻辑程序的良基语义研究

在介绍约束逻辑程序的相关概念的基础上,研究了简单单调约束逻辑程序约束原子的正文字前缀幂集展开方法,并证明展开后的正规逻辑约束与约束逻辑程序的等价特性.分析了正规逻辑程序的交替不动点良基模型建立的原理,将简单单调约束逻辑程序等价展开为与其等价的正规逻辑程序,以求展开后的逻辑程序中的给定算子的最小不动点为切入,给出了简单单调约束逻辑程序的交替不动点的良基模型.论证了文中提出的简单单调约束逻辑程序良基模型定义的合理性,说明把约束逻辑程序转化为正规逻辑程序是可行的.     

On the Relationship Between TMS and Logic Programs

The relationship between TMS and general logic programs is an important issue in non-monotonic logic programming.In this paper,we prove that,after we translate the TMS theory into a general logic program,the TMS‘s well-founded assignment (or extension) is equivalent to the corresponding general logic program‘s stable model.It means that TMS can be completely integrated into a non-monotonic logic programming environment.     

An industrial visual inspection system that uses inductive learning

This paper presents an industrial visual inspection system that uses inductive learning. The system employs RULES-3 inductive learning algorithm to extract the necessary set of rules and template matching technique to process an image. Twenty 3×3 masks are used to represent an image. Each example consists of 20 frequencies of each mask. The system was tested on five different types of tea or water cups in order to classify the good and bad items. The system was trained using five good cups and then tested for 113 unseen examples. The results obtained showed the high performance of the system: the efficiency of the system for correctly classifying unseen examples was 100%. The system can also decide what type of the cup is being processed.