Artificial Intelligence: introduction

Stefano De Luca

Practical stuff
Stefano De Luca Mail:
deluca@mat.uniroma2.it s.deluca@ieee.org

Textbook:

Receive: after lesson, by appointment

S. Russell and P. Norvig Artificial Intelligence: A Modern Approach Prentice Hall, 2003, Second Edition In Italian: Intelligenza artificiale. Un approccio moderno, 1 vol., Pearson Education Italia,2005, 40 Exam:
written at end and oral If you ask for: intermediate written exam

AI 13/10/2008 Pag.2

Course overview
What is AI. Intelligent agents. Knowledge representation RDF, OWL Semantic web Logic and Reasoning Problem solving. Planning Uncertain knowledge and reasoning. Neural Network Genetic Algorithms Industrial case histories

AI 13/10/2008 Pag.3

Outline
What is AI A brief history The State of the art (see book)

AI 13/10/2008 Pag.4

What is Artificial Intelligence

Creative extension of philosophy:
Understand and BUILD intelligent entities

Origin after WWII Highly interdisciplinary Currently consist of huge variety of subfields
This course will discuss some of them
AI 13/10/2008 Pag.5

Old AI

AI 13/10/2008 Pag.6

Now: Smart Music System

The Bose uMusic system uses artificial intelligence to learn the listening habits and preferences of its users. Load your CDs into the digital music delivery system, it can hold thousands of songs, and it will learn your listening preferences and prioritize your music collection.
AI 13/10/2008 Pag.7

Lastfm.it, Pandora.com

AI 13/10/2008 Pag.8

Control systems

AI 13/10/2008 Pag.9

What is Artificial Intelligence

Different definitions due to different criteria
Two dimensions:
Thought processes/reasoning vs. behavior/action Success according to human standards vs. success according to an ideal concept of intelligence: rationality. Systems that think like humans Systems that act like humans Systems that think rationally Systems that act rationally

AI 13/10/2008 Pag.10

Systems that act like humans

When does a system behave intelligently?
Turing (1950) Computing Machinery and Intelligence Operational test of intelligence: imitation game

Test still relevant now, yet might be the wrong question. Requires the collaboration of major components of AI: knowledge, reasoning, language understanding, learning,

AI 13/10/2008 Pag.11

Systems that act like humans

Andrew Hodges. Alan Turing, the enigma (Storia di un Enigma, Bollati Boringhieri)

Problem with Turing test: not reproducible, constructive or amenable to mathematical analysis.
AI 13/10/2008 Pag.12

Systems that think like humans

How do humans think?
Requires scientific theories of internal brain activities (cognitive model):
Level of abstraction? (knowledge or circuitry?) Validation?
Predicting and testing human behavior Identification from neurological data

Cognitive Science vs. Cognitive neuroscience.

Both approaches are now distinct from AI Share that the available theories do not explain anything resembling human intelligence.
Three fields share a principal direction.
AI 13/10/2008 Pag.13

Systems that think like humans

Some references;
Daniel C. Dennet. Consciousness explained. M. Posner (edt.) Foundations of cognitive science Francisco J. Varela et al. The Embodied Mind J.-P. Dupuy. The mechanization of the mind

AI 13/10/2008 Pag.14

Systems that think rationally

Capturing the laws of thought
Aristotle: What are correct argument and thought processes?
Correctness depends on irrefutability of reasoning processes.

This study initiated the field of logic.

The logicist tradition in AI hopes to create intelligent systems using logic programming.

Problems:
Not all intelligence is mediated by logic behavior What is the purpose of thinking? What thought should one have?
AI 13/10/2008 Pag.15

Systems that think rationally

A reference;
Ivan Bratko, Prolog programming for artificial intelligence.

AI 13/10/2008 Pag.16

Systems that act rationally

Rational behavior: doing the right thing
The Right thing is that what is expected to maximize goal achievement given the available information.

Can include thinking, yet in service of rational action.

Action without thinking: e.g. reflexes.

AI 13/10/2008 Pag.17

Systems that act rationally

Two advantages over previous approaches:
More general than law of thoughts approach More amenable to scientific development.

Yet rationality is only applicable in ideal environments. Moreover rationality is not a very good model of reality.

AI 13/10/2008 Pag.18

Systems that act rationally

Some references

AI 13/10/2008 Pag.19

Rational agents
An agent is an entity that perceives and acts This course is about designing rational agents
An agent is a function from percept histories to actions:

f : P* A

For any given class of environments and task we seek the agent (or class of agents) with the best performance. Problem: computational limitations make perfect rationality unachievable.
AI 13/10/2008 Pag.20

Foundations of AI
Different fields have contributed to AI in the form of ideas,viewpoints and techniques.
Philosophy: Logic, reasoning, mind as a physical system, foundations of learning, language and rationality. Mathematics: Formal representation and proof algorithms, computation, (un)decidability, (in)tractability, probability. Psychology: adaptation, phenomena of perception and motor control. Economics: formal theory of rational decisions, game theory. Linguistics: knowledge represetatio, grammar. Neuroscience: physical substrate for mental activities. Control theory: homeostatic systems, stability, optimal agent design.

AI 13/10/2008 Pag.21

A brief history
What happened after WWII?
1943: Warren Mc Culloch and Walter Pitts: a model of artificial boolean neurons to perform computations.
First steps toward connectionist computation and learning (Hebbian learning). Marvin Minsky and Dann Edmonds (1951) constructed the first neural network computer

1950: Alan Turings Computing Machinery and Intelligence

First complete vision of AI. Idea of Genetic Algorithms

AI 13/10/2008 Pag.22

A brief history (2)

The birth of (the term) AI (1956)
Darmouth Workshop bringing together top minds on automata theory, neural nets and the study of intelligence.
Allen Newell and Herbert Simon: The logic theorist (first nonnumerical thinking program used for theorem proving) For the next 20 years the field was dominated by these participants.

Great expectations (1952-1969)

Newell and Simon introduced the General Problem Solver.
Imitation of human problem-solving

Arthur Samuel (1952-)investigated game playing (checkers ) with great success. John McCarthy(1958-) :
Inventor of Lisp (second-oldest high-level language) Logic oriented, Advice Taker (separation between knowledge and reasoning)

AI 13/10/2008 Pag.23

A brief history (3)

The birth of AI (1956)
Great expectations continued ..
Marvin Minsky (1958 -)
Introduction of microworlds that appear to require intelligence to solve: e.g. blocks-world. Anti-logic orientation, society of the mind.

Collapse in AI research (1966 - 1973)

Progress was slower than expected.
Unrealistic predictions.

Some systems lacked scalability.

Combinatorial explosion in search.

Fundamental limitations on techniques and representations.

Minsky and Papert (1969) Perceptrons.

AI 13/10/2008 Pag.24

A brief history (4)

AI revival through knowledge-based systems (1969-1970)
General-purpose vs. domain specific
E.g. the DENDRAL project (Buchanan et al. 1969)
First successful knowledge intensive system.

Expert systems
MYCIN to diagnose blood infections (Feigenbaum et al.)
Introduction of uncertainty in reasoning.

Increase in knowledge representation research.

Logic, frames, semantic nets,

AI 13/10/2008 Pag.25

A brief history (5)

AI becomes an industry (1980 - present)
R1 at DEC (McDermott, 1982) Fifth generation project in Japan (1981) American response

Puts an end to the AI winter. Connectionist revival (1986 - present)

Parallel distributed processing (RumelHart and McClelland, 1986); back-propagation

AI 13/10/2008 Pag.26

A brief history (6)

AI becomes a science (1987 - present)
Neats vs. scruffies.
In speech recognition: hidden markov models In neural networks In uncertain reasoning and expert systems: Bayesian network formalism

The emergence of intelligent agents (1995 - present)

The whole agent problem:
How does an agent act/behave embedded in real environments with continuous sensory inputs
AI 13/10/2008 Pag.27

The Hype Cycle of Emerging Technologies (Gartner 2005)

AI 13/10/2008 Pag.28

Major research areas

(Applications)

Natural Language Understanding Image, Speech and pattern recognition Uncertainty Modeling Expert systems Virtual Reality ..
AI 13/10/2008 Pag.29

Symbolic Programming
Program as Representation of world Symbol as basic element of representation
atom, property, relationship

Symbolic Expression as method of combination LISP for Symbolic programming PROLOG for logic programming Object-Oriented Concept

Pag.

Knowledge Representation
What kind of Knowledge needed for Problem solving ? Structure of knowledge ?
declarative vs procedural

Representation techniques ?
explicit vs (implicit + inference) logic, frame, object-oriented, semantic net, script

Knowledge acquisition and update

Pag.

Search Theory
An Optimization method Analyze alternative cases and select one Cope with Exponential complexity, NP classes Genetic Algorithm, Simulated Annealing
Stochastic search Try likely one first (Heuristic Search) Utilize local information (Hill Climbing Method) Optimal solution vs good solution

Pag.

Automated Reasoning
Qualitative Reasoning
Ostrich flys ?

Utilization of qualitative knowledge such as

Non-monotonic Reasoning Plausible Reasoning

Utilization of Experience
Pag.

Information fusion under uncertainty

Case-based Reasoning

Machine Learning
Performance improvement by experience
How much of knowledge required to start learning ? Method of acquiring new knowledge and merging it to existing knowledge-base Role of teacher Role of examples and experience

Parameter Adjustment Inductive learning Computational Learning Theory

Used in Practice such as Data Mining

Pag.

Quality of generalization capability in terms of Training data

Data Mining
Knowledge extraction for decision making

Data

Information / knowledge

Decision
Making

AI 13/10/2008 Pag.35

Neural Network Computational model of Neurons

Power comes from Connection of simple processing element connectionism
X1 X2 . . . Xn w1 w2 wn

F(X1, X2, , Xn)

Pag.

Neural Network
learning = link weigh adjustment
Error-back-propagation : supervised learning Any Functional Mapping is learnable

Strong at Sensory Data Processing

Symbolic Grounding

Old Horse on the race again

Massive parallelism, graceful degradation
AI 13/10/2008 Pag.37

Neural Network Classifier

Job(1/0) age Salary #mouth Debt good

medium

bad

Input layer
AI 13/10/2008 Pag.38

Hidden layer

Output layer

Genetic Algorithms
Computational model of life evolution Stochastic optimization technique
Initial chromosome creation New generations are made (cross over, mutation) survival of the fittest study evolution of life, by simulation
Pag.

Base of artificial life research

AI Success Story (Planning)

MARVEL
(Schwuttke, 1992)

Berth assignment (KAL, 1997) Unmanned Vehicle Pathfinder Rover, 1996 Asimo a walking robot
Ground and air

Real-time Space shuttle Mission planning

AI 13/10/2008 Pag.40

Autonomous Land Vehicle

(DARPAs GrandChallenge contest)

AI 13/10/2008 Pag.41

systems

AI Success Story : Medical expert

Programs listed by Special Field
Antibiotics & Infectious Diseases Cancer Chest pain Dentistry Dermatology Drugs & Toxicology Emergency Epilepsy Family Practice Genetics Geriatrics Gynecology Imaging Analysis Internal Medicine Intensive Care Laboratory Systems Orthopedics Pediatrics Pulmonology & Ventilation Surgery & Post-Operative Care Trauma Management

AI 13/10/2008 Pag.42

Pattern Recognition Applications

Handwriting and document recognition Signature, biometrics (finger, face, iris, etc.) Gesture, facial expression EEG, EKG, X-ray Trafic monitoring, Remote Sensing Smart Weapon guided missile, target homing
AI 13/10/2008 Pag.43

forms, postal mail, historic documents PDA pen recognition

As a Human computer intertraction

Handwriting Understanding

AI 13/10/2008 Pag.44

Decision Support Systems (DSS)

AI 13/10/2008 Pag.45

Intelligent Transportation Systems

AI 13/10/2008 Pag.46

Future of AI
Making AI Easy to use
Easy-to-use Expert system building tools Web auto translation system Recognition-based Interface Packages

Integrated Paradigm
Symbolic Processing + Neural Processing

AI in everywhere, AI in nowhere
AI embedded in all products Ubiquitous Computing, Pervasive Computing

AI 13/10/2008 Pag.47

Other courses at the UniRoma2

AI does not end here

Sistemi di agenti (Specialistica) Cooperazione di Agenti informatici (Specialistica)

AI 13/10/2008 Pag.48

Some references
Understanding Intelligence by Rolf Pfeifer and Christian Scheier. Artificial Intelligence: Structures and Strategies for Complex Problem-solving by George Luger. Computation and Intelligence: Collective readings edited by George Luger. Paradigms of Artificial Intelligence Programming: Case Studies in Common Lisp by Peter Norvig.
AI 13/10/2008 Pag.49