Lecture 5_Learning set of rules 机器学习概论 教程文件.pptVIP

Lecture 5:Learning Rule;回顾－1;回顾－2;回顾－3;Homework;Decision-Trees to Rules;顺序覆盖算法;SC Algorithm - An Example;SC Algorithm - An Example;SC Algorithm - An Example;SC Algorithm - An Example;SC Algorithm - An Example;SC Algorithm - An Example;SC Algorithm - An Example;No-optimal Covering Example;No-optimal Covering Example;No-optimal Covering Example;No-optimal Covering Example;No-optimal Covering Example;No-optimal Covering Example;No-optimal Covering Example;No-optimal Covering Example;No-optimal Covering Example;学习单个规则 (Learn one rule)的策略;Top-Down Rule Learning - An Example;Top-Down Rule Learning - An Example;Top-Down Rule Learning - An Example;Top-Down Rule Learning - An Example;Top-Down Rule Learning - An Example;Bottom-Up Rule Learning - An Example;Bottom-Up Rule Learning - An Example;Bottom-Up Rule Learning - An Example;Bottom-Up Rule Learning - An Example;Bottom-Up Rule Learning - An Example;Bottom-Up Rule Learning Example;Bottom-Up Rule Learning - An Example;Bottom-Up Rule Learning - An Example;Bottom-Up Rule Learning - An Example;Bottom-Up Rule Learning - An Example;Bottom-Up Rule Learning - An Example;Learn-One-Rule– An Top-Down Method;可采用beam search Beam search :始终维护一个当前k个性能最好的hypotheses所构成的candidate- hypotheses。由这些候选的hypotheses各自生成的candidate-literals构成一个大的candidate-literals，并在这个大的candidate-literals找出最好的best-literals。 部分解决贪心法的局部最优的缺点。 很容易推广大多值目标概念（函数） 评估函数 (Performance) ;Rule Pruning;An Example;学习规则：小结;规则学习－问题;学习一阶谓词规则;ILP Systems;FOIL: --- An Algorithm for Learning Sets of First-Order Rules;FOIL: Description;Foil-Gain 度量;FOIL中规则的特化;FOIL – An Example;FOIL - Negative Training Data;FOIL Induction;FOIL Induction;FOIL Induction;FOIL Induction;FOIL Induction;FOIL Induction;FOIL Induction;FOIL Induction;FOIL Induction;FOIL Induction;FOIL Induction;FOIL Induction;FOIL Induction;FOIL Induction;FOIL Induction;FOIL－局限;Neg: {1,1,1,4,2,1,2,2,2,3,2,4,2,5,2,6, 3,1,3,2,3,3,3,4,4,1,4,3,4,4,5,1, 5,2,5,3,5,4,5,5,5,6,6,1,6,2,6,3, 6,4,6,6};下一讲