人工智能导论知识点总结一

人工智能导论知识点总结一

Knowledge Representation and Reasoning

  1. 专有名词归纳整理

专有名词

翻译

专有名词

翻译

Semantic Networks

语义网络

Reification

实体化

Abduction

溯因推导

Deduction

演绎

Induction

归纳

Monotonic

单调

 

2.【考点/客观题】常见的Deductive/logical methods【演绎/逻辑方法】

(1)Forward-chaining production rule systems【前向链式生产规则系统】

(2)Semantic networks【语义网络】 (3)Frame-based systems【框架系统】

(4)Description logics【逻辑描述】

3. 【考点/客观题】常见的Abductive/uncertain methods【溯因/不确定方法】

(1)Bayesian reasoning【贝叶斯推理】 (2)Default reasoning【默认推理】

(3)rule-based methods【基于规则方法】(4)Dempster-Shafer theory【D-S证据理论】

(5)fuzzy reasoning【模糊推理】

4. 【考点/客观题】语义网络概念:A semantic network is a simple representation scheme【一种简单表示方案】 that uses a graph of labeled nodes and labeled, directed arcs to encode knowledge.

5. 【考点/客观题】ISA与AKO

(1)ISA: link instances 【实例】to classes【类】

(2)AKO:classes【类】 to superclasses 【超类】

6. 【考点/客观题】连接类型

人工智能导论知识点总结一_第1张图片

 7. 【考点/客观题】实体化的概念:non-binary relationships 【非二进制的关系】can be represented by “turning the relationship into an object”

8.【考点/客观题】溯因推理相关内容

(1)溯因:a reasoning process that tries to form plausible 【看似合理的】explanations for observations

注意:从本质上来说溯因是unsound and uncertain【不可靠和不确定的】!!!

(2)Abductive reasoning【溯因推理】:Reasoning that derives an explanatory hypothesis from a given set of facts 【从一组给定的事实中推导出解释性假设的推理】

  • 特点:结论是假设并不是定理,也许有多种貌似合理的假设,Reasoning as a hypothesize-and-test cycle 【作为假设-测试循环的推理】
  • 应用:

①word sense disambiguation in natural language process【词义消歧】

②image understanding【图像理解】  ③criminal investigation【刑事侦查】

④medical diagnosis 【医疗诊断】

9.【考点/客观题/主观题(简答题)】Three different logic-based reasoning systems:

abduction【溯因】/ deduction【演绎】/induction【归纳】比较

abduction

(abductive reasoning)

Deduction

(deductive reasoning)

Induction

(inductive reasoning)

reasons from causes to effects

【从结果倒推原因】

reasons from effects to causes

【原因推导出结果】

reasons from specific cases to general rules

【特殊情况推广到一般规律】

non-monotonic【非单调性】

monotonic【单调】

non-Monotonic【非单调】

uncertain 【不确定的】

Certain【确定的】

uncertain 【不确定的】

可信度会随着新的事实出现增加或者减弱

绝不会改变一个句子真值

许多假设也许是会被丢弃的

注意:演绎推理的结论一定正确!!!

10.【考点/客观题】不确定的来源

(1) Input  (2)Knowledge  (3)Output

11.【考点/客观题】uncertainty representations

  • Evidential reasoning
  • Fuzzy reasoning
  • Default reasoning 【缺省推理】
  • Rule-based methods

14.  【考点/客观题】Uncertainty tradeoffs【不确定性的权衡】

  • Bayesian networks
  • Nonmonotonic logic
  • Certainty factors
  • Dempster-Shafer theory
  • Fuzzy reasoning

你可能感兴趣的:(计算机,人工智能)