Introduction to Artificial Intelligence

Lê Anh Cường
Faculty of Information Technology
Ton Duc Thang University
Ho Chi Minh city, Viet Nam
 Outline
1.AI systems
2.Architecture of an AI system
3.Problems
 Watson IBM
●Watson is a question answering computer system capable of answering questions posed
in natural language,developed in IBM.
●In 2011, Watson competed on Jeopardy! against former winners Brad Rutter and Ken
Jennings.Watson received the first place prize of $1 million.
Watson IBM
Watson IBM
 Game (Deep Blue)
Deep Blue was a chess-playing computer developed by IBM.
●

Deep Blue won its first game against a world champion on February 10, 1996,
●

when it defeated Garry Kasparov.

 Game (AlphaGo)
●AlphaGo is a computer program developed by Google DeepMind. In
October 2015, the first Computer Go program to beat a professional
human. In March 2016, it beat Lee Sedol.
 Self Driving Cars
Google selt-drving cars
●

●Elon Musk is the founder, CEO, and CTO of SpaceX; co-founder, CEO, and
product architect of Tesla Motors; co-founder and chairman of SolarCity; co-
chairman of OpenAI;
Google Machine Translation
Model Translation
Translation statistical machine translation System
material
Cortana-Microsoft, Siri-Apple
 Facebook
Smart advertisment
●

Smart friends connection

●
 Outline
1.AI systems
2.Architecture of an AI system
3.Problems
KB Systems Architecture
Knowledge Base
 Outline
1.AI systems
2.Architecture of an AI system
3.Problems
 Knowledge-based Systems: a definition

●A system that draws upon the knowledge of human

experts captured in a knowledge base to solve
problems that normally require human expertise.
● Properties:
– Heuristic rather than algorithmic
– Highly specific domain knowlege

Knowledge Based System = Knowlege Base + Inference Engine

 Knowledge Representation & Inference

Knowledge Representation Inference

Logic
● Resolution principle
●

Production rules
● Backward: top-down & goal directed
●

Forward: bottom-up & data driven

●

Semantic nets & frames

● Inheritance & advanced reasoning
●

Case-based reasoning
● Similarity based
●
 Some KSBs
● DENDRAL (chemical): 1965
● MYCIN (medicine): 1970
● XCON/RI (computer): 1981
● DRILLING Advisor (industry): 1983
● Human Resource Management
 Tasks of KBSs
● Diagnosis
● Interpretation
● Prediction
● Design
● Plan
● Monitoring
● Control
● Instruction
● Debugging
 Advantages of KBSs
● Increase availability of expert knowledge
– Expertise not accessible
– Training future experts
● Efficient and cost effective
● Consistency of answers
● Explanation of solution
● Deal with uncertainty
 Limitations
● Lack of common sense
● Inflexible, difficult to modify
● Restricted domain of expertises
● Lack of learning ability
● Not always reliable
What is AI?
Views of AI fall into four categories

Thinking humanly Thinking rationally

Acting humanly Acting rationally

Thinking humanly: cognitive
modeling
1960s "cognitive revolution": information processing psychology
Requires scientific theories of internal activities of the brain
How to validate? Requires
1) Predicting and testing behavior of human subjects (topdown)
or 2) Direct identification from neurological data (bottomup)

Both approaches (roughly, Cognitive Science and Cognitive

Neuroscience) are now distinct from AI
Thinking rationally: “laws of
thought”
Aristotle: what are correct arguments/thought processes?
Several Greek schools developed various forms of logic: notation
and rules of derivation for thoughts; may or may not have
proceeded to the idea of mechanization
Direct line through mathematics and philosophy to modern AI
Problems:
1. Not all intelligent behavior is mediated by logical deliberation
2. What is the purpose of thinking? What thoughts should I have?
Acting Humanly: Turing Test
Acting rationally: rational agent
Rational behavior: doing the right thing
The right thing: that which is expected to maximize goal
achievement, given the available information
Doesn't necessarily involve thinking – e.g., blinking reflex – but
thinking should be in the service of rational action
Rational agents
An agent is an entity that perceives and acts
This course is about designing rational agents
Abstractly, an agent is a function from percept histories to
actions:
[f: P* -> A]

For any given class of environments and tasks, we seek the agent
(or class of agents) with the
best performance
Caveat: computational limitations make perfect rationality
unachievable design best program for given machine resources
AI prehistory
Philosophy Logic, methods of reasoning, mind as physical system
foundations of learning, language, Rationality

Mathematics Formal representation and proof algorithms, computation,

(un)decidability, (in)tractability, probability

Economics utility, decision theory

Neuroscience physical substrate for mental activity
Psychology phenomena of perception and motor control,
experimental techniques
Computer building fast computers
engineering
Control theory design systems that maximize an objective
function over time
Linguistics knowledge representation, grammar
State of the art
Robotic vehicles can drive autonomously in most situations (95%). CMU’s
BOSS can drive through an urban environment, following traffic rules and
avoiding pedestrians.
Customers can call United Airlines to book flights, or use Google Voice to
translate their native speech into other languages
Deep Blue beat the world-reigning chessmaster Gary Kasparov in 1997, and
computers have continued to convincingly beat humans in recent years.
Learning algorithms help to classify spam mail, helping all email users save
time, sorting out over 80-90% of mail as spam traffic.
Both military and commercial sectors employ AI to handle logistics.Aircraft
routing and convoy logistics A.I. are used to coordinate the movement of
massive numbers of supplies and units according to constraints
Agents
An agent is anything that can be viewed as perceiving its
environment through sensors and acting upon that environment
through actuators

Human agent: eyes, ears, and other organs for sensors; hands,
legs, mouth, and other body parts for actuators

Robotic agent: cameras and infrared range finders for sensors;

various motors for actuators

2019/8/15 503043-INTRODUCTION TO AI & INTELLIGENT AGENTS 32

Agents and environments

The agent function maps from percept histories to actions:

[f: P* -> A]
The agent program runs on the physical architecture to produce f
Agent = architechture + program

2019/8/15 503043-INTRODUCTION TO AI & INTELLIGENT AGENTS 33

Vacuum-learner word
Percepts: location and contents, e.g.,
[A,Dirty]
Actions: Left, Right, Suck, NoOp

2019/8/15 503043-INTRODUCTION TO AI & INTELLIGENT AGENTS 34

A vacuum-learner agent

2019/8/15 503043-INTRODUCTION TO AI & INTELLIGENT AGENTS 35

Rational agents
An agent is an entity that perceives and acts
This course is about designing rational agents
Abstractly, an agent is a function from percept histories to
actions:
[f: P* -> A]

For any given class of environments and tasks, we seek the agent
(or class of agents) with the
best performance
Caveat: computational limitations make perfect rationality
unachievable design best program for given machine resources
PEAS
PEAS: Performance measure, Environment, Actuators, Sensors
Must first specify the setting for intelligent agent design
Consider, e.g., the task of designing an automated taxi driver:
oPerformance measure
oEnvironment
oActuators
oSensors

2019/8/15 503043-INTRODUCTION TO AI & INTELLIGENT AGENTS 37

Environment types
Fully observable (vs. partially observable): An agent's sensors give it access to
the complete state of the environment at each point in time.
Deterministic (vs. stochastic): The next state of the environment is completely
determined by the current state and the action executed by the agent. (If the
environment is deterministic except for the actions of other agents, then the
environment is strategic)
Episodic (vs. sequential): The agent's experience is divided into atomic
"episodes" (each episode consists of the agent perceiving and then performing a
single action), and the choice of action in each episode depends only on the
episode itself.

2019/8/15 503043-INTRODUCTION TO AI & INTELLIGENT AGENTS 38

Environment types
Fully observable (vs. partially observable): An agent's sensors give it access to
the complete state of the environment at each point in time.
Deterministic (vs. stochastic): The next state of the environment is completely
determined by the current state and the action executed by the agent. (If the
environment is deterministic except for the actions of other agents, then the
environment is strategic)
Episodic (vs. sequential): The agent's experience is divided into atomic
"episodes" (each episode consists of the agent perceiving and then performing a
single action), and the choice of action in each episode depends only on the
episode itself.

2019/8/15 503043-INTRODUCTION TO AI & INTELLIGENT AGENTS 39

Agent functions and programs
An agent is completely specified by the agent function mapping
percept sequences to actions
One agent function (or a small equivalence class) is rational
Aim: find a way to implement the rational agent function
concisely

2019/8/15 503043-INTRODUCTION TO AI & INTELLIGENT AGENTS 40

Table-lookup agent

Drawbacks:
oHuge table
oTake a long time to build the table
oNo autonomy
oEven with learning, need a long time to learn the table entries

2019/8/15 503043-INTRODUCTION TO AI & INTELLIGENT AGENTS 41

Agent program for a vacuum-
cleaner agent

2019/8/15 503043-INTRODUCTION TO AI & INTELLIGENT AGENTS 42

Agent types
Four basic types in order of increasing generality:

oSimple reflex agents

oModel-based reflex agents
oGoal-based agents
oUtility-based agents

2019/8/15 503043-INTRODUCTION TO AI & INTELLIGENT AGENTS 43

Simple reflex agents

2019/8/15 503043-INTRODUCTION TO AI & INTELLIGENT AGENTS 44

Simple reflex agents

2019/8/15 503043-INTRODUCTION TO AI & INTELLIGENT AGENTS 45

Model-based reflex agents

2019/8/15 503043-INTRODUCTION TO AI & INTELLIGENT AGENTS 46

Goal-based agents

2019/8/15 503043-INTRODUCTION TO AI & INTELLIGENT AGENTS 47

Utility-based agents

2019/8/15 503043-INTRODUCTION TO AI & INTELLIGENT AGENTS 48

Learning agents

2019/8/15 503043-INTRODUCTION TO AI & INTELLIGENT AGENTS 49