【人工智能】人工智能导论

什么是人工智能

人工智能：用人工的方法在机器上实现的智能

人工智能学科

智能如何产生

Theory of mind:理论思维是智慧的本质,智能由思维生成—逻辑推理

Knowledge-threshold-based theory:intelligent depends on the quantity of knowledge and the level of generalizability 知识阈值理论—搜索思想

进化理论；将知识的表现形式替换为控制

智能特点

感知能力

Perception:the capacity to sense the outside world by sensory organs

视觉

听觉

记忆和思维能力

逻辑思维logical thinking(abstract thinking)

Like:三段论

Think logical

The thought process is serial

Easy to formalize

Thought process with rigor,reliability

形象思维Image thinking(direct thinking)

Follow your intuition

Thought processes are

顿悟，灵感Insight thinking(inspired thinking)

Sudden bursts at irregular intervals

学习能力

有指导学习

无指导学习

行为能力

能够表示出来

图灵测试

图灵机

人工智能的研究内容

知识表示—知识形式化和建模

knowledge representation:a formalization and modeling of human knowledge

符号表示

一阶谓词逻辑

产生式系统

连接机制表示

人工智能网络（深度学习）

机器感知

查看上文【感知能力】

机器思维

通过对信息的处理（机器感知）实现

机器学习

深度学习

Deep learning: It is a hot topic in current research and application.

Deep learning focuses on constructing neural networks that mirror the functionality of human brains. These multi-layered neural networks, akin to the human brain, are capable of gathering information and generating corresponding behavior.

SVM：支持向量机

机器行为

研究如何让计算机说，写，画

研究领域

智能信息检索intelligent information retrieval

能理解自然语言

具有推理能力

有一定常识性知识

数据挖掘与知识发现Data mining and knowledge discovery

A primary goal of data mining is to discover meaningful patterns within the data collection framework. These patterns may include rule sets, clusters, decision trees, and knowledge represented by a dependent network.

Data mining processes:

数据预处理

建模

模型评估以及模型应用

专家系统Expert System

The expert system imitates the thought patterns of human experts to address issues within their specialized area, achieving a level that is on par with or exceeds that of human experts.

自动编程Automatic programming

Program synthesis: Program synthesis requires users to instruct computers to perform specific tasks, such as "do what" or "automate repetitive processes", without detailing how it should be done. Computers are capable of automatically realizing the design through program synthesis techniques.

Verification of program correctness involves identifying and establishing a systematic collection of theories and methodologies. By utilizing these theories and methodologies, one can effectively utilize them to verify the program's correctness.

机器人Robots

Robotic advancement encompasses programmable robots (1st generation), adaptive robots (2nd generation), and contemporary intelligently designed robots.

Unimet

Voshatlan

组合优化Combinatorial optimization problem

NP-Complete problems require a significant amount of computational effort when solved using the most efficient approaches, as they grow significantly with increasing problem size.

Combinatorial optimization problems

人工神经网络Artificial neural network

mp神经网络—感知器—bp神经网络—CNN神经网络

Deep neural network: the current research and application hotspot.

Deep neural network is a multi-level neural network.

Deep neural networks are guided by patterns similar to those observed in the human brain and can produce corresponding behaviors.

These neural networks have the ability to gradually perceive and comprehend objects and sounds, which means they can understand or know these objects and sounds.

分布式的AI技术与multi-agent系统之间存在密切的关系

Distributed AI systems adopt robustness as their primary criterion for ensuring system effectiveness and achieve interoperability. These systems are designed in such a way that they can seamlessly interact with diverse, non-uniform systems within dynamic and ever-evolving settings.

智能控制Intelligent control

Two salient features of intelligent control are:

Firstly, intelligent controlling represents a control mechanism integrating both non-mathematical generalized world-models and traditional mathematical models incorporating knowledge representations.

Second, the central aspect of intelligent control is high-level control. Its function is to coordinate the operational environment and processes—essentially managing decision-making and planning—aimed at achieving generalized problem-solving capabilities.

The basic types of intelligent control are:
(1) expert intelligent control

(2) 模糊控制fuzzy control

(3) neural network control

智能仿真Intelligent simulation

Intelligent-based simulation represents the integration of AI technology into the field of simulation, resulting in its formation as a new type of system.

通过动态模型的实验来实现行为生成器驱动模型在指定条件下产生行为。模拟旨在基于描述性知识、目的性知识以及处理性知识得出结论性的知识。

智能CADIntelligent CAD

Intelligently Computer-Aided Design (ICAD) refers to integrating advanced artificial intelligence technologies into computer-aided design workflows and establishing intelligent CAD systems. AI can be effectively integrated into diverse areas of CAD technology. Taking into account the current technological trends, it is anticipated that this emerging field will encompass a variety of prominent areas, including but not limited to optimization techniques, simulation enhancements, user interface advancements, and data-driven decision-making capabilities.

(1)设计自动化 Design automation.

(2) Intelligent interaction.

(3) Intelligent graphics.

(4) Automatic data acquisition

智能CAI Intelligent CAI

AI辅助教学系统旨在引入人工智能技术到计算机辅助教学领域。ICAI系统通常被划分为专业知识模块、教学策略以及学生模型相关的智能接口和自然语言处理相关的智能接口。

ICAI should have the following intelligence characteristics:

(1) Automatically generate questions and exercises.

(2) 自动选择并根据需要调整教学内容与进度

students' learning situation.

Without human intervention, automatically address and resolve issues, while simultaneously producing responses based on a thorough comprehension of the underlying principles and requirements.

teaching content.

(4) Ability to generate and understand naturallanguage.

(5) Have the ability to understand and consult the teaching content.

能够识别学生的问题并深入分析原因并采取相应的补救措施

(7) Able to evaluate students' learning behavior.

(8) Can continuously improve teaching strategiesin teaching.

智能管理与智能决策：Intelli gent管理与智能化决策支持系统

Intelligent management involves integrating artificial intelligence technologies into managerial practices. It entails establishing intelligent management systems and investigating strategies for enhancing the intelligence quotient (IQ) of computer-based systems. Additionally, it focuses on developing theoretical frameworks, methodologies, and implementation approaches for designing intelligent systems.

The implementation of intelligent decisions involves integrating artificial intelligence technology into the decision-making framework and establishing intelligent decision support systems.

智能多媒体Intelligent multimedia system

A multimedia computer system is a computer system that has the ability to efficiently handle various types of media information, including textual content, visual data, pictorial representations, and audio information.

Intellient Multimedia aims to incorporate advanced artificial intelligence technologies into the field of multimedia systems, enabling its functionalities and operational capabilities to be further enhanced in terms of functionality and performance.

智能操作系统Intelligent operating system

Intelligent computer system

智能通信Intelligent Communication

智能网络系统Intelligent network system

人工生命Artificial Life

知识表示knowledge representation

什么是知识

知识的特征

不确定性uncertainty

可表示性和可应用性

知识分类

scope

常识

专业知识

按功能性分类

事实性知识

控制性知识

过程性知识

知识表示

人类知识的形式化与模型化

一组转换规则的集合

状态空间表示State space

状态：n元组

操作符（算子）

1.设定初始状态

2.设定目标状态

3.经过一系列操作后，将初始状态转变为目标状态

15数码问题

在4x4的棋盘里放入15个数字，一个数码排列方式代表一个状态

操作最多有四个，但避免还原，需要去掉一个操作

状态图：有向图，状态用节点来表示，有向边表示操作

问题规约表示法

把问题分解为原子性的子问题

规约图

一阶谓词逻辑

产生式系统

The control system does the following :

Retrieve relevant information from within the rule base and be matched against the existing body of knowledge stored within this comprehensive database, thereby identifying corresponding known facts.

Several rules can be successfully matched, and conflict resolution is carried out.

当运行一条特定指令时,若其右侧为一个或多个结果,则将这些结果加入综合数据库;若右侧为一个或多个运算,则依次执行这些运算.

面对不确定性知识，在执行每一条规则时，应计算出结论的不确定性，并按照特定算法加以确定。

Verify if the final conclusion is included in the comprehensive database, and decide whether to halt system operation.

框架语意网络

搜索和求解策略

搜索

盲目搜索

回溯

深度优先

宽度优先搜索

启发式搜索

按运用的方法分类:

1)陈述性启发信息:用于更准确、更精炼地描述状态

2)过程性启发信息:用于构造操作算子

3)控制性启发信息:表示控制策略的知识

按作用分类:

1)用于扩展节点的选择,即用于决定应先扩展哪一个节点,以免盲目扩展。

用于决定生成哪些后续节点的选择, 即为避免由于随意或不计划地生成大量无意义的后续节点而导致的问题

选择用于删除节点的方法,即用来确定应舍弃的无用节点,避免产生额外的时间和空间消耗