人工智能英文版課件：01 intoduction

1、Artificial IntelligenceCourse OutlineGrading SchemeAttendance (10%)Assignment (20%)2 assignmentsMid-term Test (25%)Around 9th weekFinal Exam (45%)Scope of 2013 Artificial IntelligenceIntroduction of AI (Chapter 1)Intelligent Agents (2)Solving Problem by Searching (3)Beyond Classical Search (4)Advers

2、arial Search (5)Constraint Satisfaction Problems (6)Learning from Examples (18)Planning (10)Current Research Topics (Pending)Dealing with Big Data, e.g. Image RetrievalDeep Learning for Image ClassificationChapter 1 Introduction of Artificial intelligenceOutlineWhy we study AIWhat is AI?The Foundati

3、ons of AIThe History of artifical intelligenceThe State of ArtSummaryWhy we study AIHuman a.k.a. Homo Sapiens: Latin of Wise HumanMental Capacities learn, predict, create, etcSense of SelfAI address one of the ultimate puzzlesHow is it possible for a slow, tiny brain to perceive, understand, predict

4、 and manipulate a world far larger and more complicated than itself?Artificial Intelligence or AITry to understand intelligent entitiesTry to build intelligent entities (unlike philosophy & psychology)A relative new research area started in 1956With a large influence to many other research areasPatt

5、ern Recognition, Machine Learning, Robotics, Controls, Man-Machine Interface, etcWhy we study AIIntelligent RobotsDancingPlaying footballR2-D2 and 3CPOand?!Why we study AIIntelligent Game PlayingAI in GamesWhy we study AIPattern recognitionImage ClassificationStock Market Candlestick PatternsHandwri

6、tten charactersand Why we study AIIntelligent Traffic ControlWhy we study AISearch enginesWhy we study AIGoogle TranslationTranslate whole paragraphPowerfulreally?What is AI?There is no solid definition of Artificial, the definition differs for different peopledifferent contextsdifferent historical

7、periods. Some definitions of AI organized into four categories 1.Systems that think like humans. 2.Systems that think rationally. 3.Systems that act like humans. 4.Systems that act rationally. What is AI?The exciting new effort to make computers thinks machine with minds, in the full and literal sen

8、se” (Haugeland 1985)“The art of creating machines that perform functions that require intelligence when performed by people” (Kurzweil, 1990)“The study of mental faculties through the use of computational models” (Charniak et al. 1985)A field of study that seeks to explain and emulate intelligent be

9、havior in terms of computational processes” (Schalkol, 1990)Systems that think like humansSystems that think rationallySystems that act like humansSystems that act rationallyWhat is AI?The four definitions above vary along two dimensions :Human-centeredRationalityHuman-centered approachempirical sci

10、ence, involving hypothesis and experimental confirmation.Rationalist approachcombination of mathematics and engineering.What is AI?Acting humanly: Turing Test Alan Turings 1950 article Computing Machinery and Intelligence proposed Turing Test,which provide a satisfactory operational definition of in

11、telligence. Computing Machinery and IntelligenceWhat is AI? Turing TestWhat is AI? Computing Machinery and Intelligence“Can machines think?” “Can machines behave intelligently?The Turing test (The Imitation Game): A human interrogator poses written questions to itThe computer return written response

12、sThe computer passed the test if the human interrogator can not tell whether the responds are from a person or notWhat is AI?Computer needs to possess: Natural language processingKnowledge representationAutomated reasoningMachine learning. Turing test avoided direct physical interaction between the

13、interrogator and the computer.Disadvantages of the Turing test not reproduciblenot constructivenot amenable to mathematical analysis.What is AI?AI researchers have devoted little effort to passing the Turing testBut devoted much effort to study the underlying principle of intelligence.For example, t

14、o build a flying machine, we do not need to imitate birds but go to learn aerodynamics.“artifical flight” succeeded. Artificial versus Natural FlightWhat is AI? What does the C-3PO need to communicate with human?What is AI?What does the C-3PO need to communicate with human? Natural language processi

15、ng to enable it to comminicate successfully in English.Knowledge representation to store what it knows or hears.Automated reasoning to use the stored information to answer questions and to draw new conclusions.What is AI?Machine learning to adapt to new circumstances and to detect and extrapolate pa

16、tterns.Computer Vision to recognize the interrogators actions and various objects presented by the interrogator.Robotics to manipulate objects and move aboutWhat is AI?Thinking Humanly: Cognitive modelingHow humans think? To determine how humans think, we need to get inside the actual working mechan

17、ism of human mind.There are two way to do this:Through introspection Through psychological experiment A program behave humanly If a programs input/output and timing behaviors match corresponding human behaviorsWhat is AI?The Neural computer is able to imitate human brainWhat is AI?1960 “Cognitive Re

18、volution”information-processing psychology replaced prevailing orthodoxy behaviorism.Requires scientific theories of internal activities of the brainWhat level of abstraction? “Knowledge” or “Circuits”?How to validate models?Predicting and testing behavior of human subjects (top-down)Direct identifi

19、cation from neurological data (bottom-up)Building computer/machine simulated models and reproduce results (simulation)What is AI?Both approaches(roughly,Cognitive Science and Cognitive Neuroscience) are now distinct from AI.Both share with AI the following characteristic:the available theories do no

20、t explain (or engender)anything resembling human-level general intelligence Hence, all three fields share one principle direction!What is AI?Thinking Rationally: Laws of Thought Aristotle: What are correct arguments/thoughts processes? codify “right thinking”-a reasoning processes Aristotles famous

21、syllogism provide patterns for argument structures. “Socrates is a man, all men are mortal; therefore Socrates is mortal” Studying the law of thought initiated the field called logic What is AI?A famous Greek philosopher: AristotleWhat is AI?Several Greek schools developed various forms of logic:not

22、ation and rules of derivation for thoughts.may or may not have proceeded to the idea of mechanizationAll kinds of things in the world and their relations can be stated by a precise notation.By 1965, programs existed that could, given enough time and memory, take a description of a problem in logical

23、 notation and find the solution to the problem if one existsLogicist tradition within AI hopes to build on such programs to create intelligent systems What is AI?problemsNot all facts are cerntainPrinciple sometimes disagrees with practiceWhat is AI?Acting Rationally: The Rational Agent 1.Acting rat

24、ionally Rational behavior: doing the right thing The right thing: is expected to maximize goal achievement for given and available information What is AI?Does not necessarily involve thinkingblinking reflex-but thinking should be in the service of rational action. Aristotle (Nicomachean Ethics)Every

25、 art and every inquiry, and similarly every action and pursuit, is thought to aim at some goodWhat is AI? 2. Rational agentsAn agent Commonly an agent is something that acts. Abstractly, an agent is a function from percept histories to actions: .f:p*-A But an computer agent is an entity that have ot

26、her attributes, such as operating under autonomous control, perceiving their environment, etcWhat is AI?Rational agentsIt is one that acts so as to achieve either the best outcome or the best expected outcome when there is uncertaintyFor any given class of environments and tasks, we seek the agent (

27、or class of agents) with the best performanceComputational limitations make perfect rationality unachievableWhat we can do is to design the best program for given machine resources.The Foundations of AIAI is the interdisciplinary study of science includingPsychologyPhilosophyNeuroscienceMathematicsL

28、inguisticsThey contributed ideas, viewpoints and techniques to AI. They are the foundations of AI.We organize these foudations around questions related to AI and contributed methods,results to AI. Operations researchControl theoryCyberneticsEconomicsComputer EngineeringThe Foundations of AIPhilosoph

29、y(428B.C.-present) Questions Can formal rules be used to draw valid conclusions? How does the mental mind arise from a physical brain? Where does knowledge come from? How does knowledge lead to action? Contributed methods and results Logic, methods of reasoning, Mind as physical system Foundations o

30、f learning, language, rationalityThe Foundations of AIMathematics(c.800- present) Questions what are the formal rules to draw valid conclusion? what can be computed? How do we reason with uncertain information? Contributed methods and results Formal representation and proof Algorithms Computation, (

31、un)decidability, (in)tractability, probabilityThe Foundations of AIEconomics(1776- present) Questions How should we make decisions so as to maximize payoff? How should we do this when others may not go along? How should we do this when the payoff may be far in the future? Contributed methods and res

32、ults utility decision theoryThe Foundations of AINeuroscience(1861- present) Questions How do brain process information? Contributed methods and results physical substrate for mental activityThe Foundations of AIPsychology(1879- present) Questions How do humans and animals think and act? Contributed

33、 methods and results phenomena of perception motor control experimental techniquesThe Foundations of AIControl theory and Cyberneties(1948- present) Questions How can we build an efficient computer? Contributed methods and results design systems that maximize an objective function over timeThe Found

34、ations of AILinguistics(1957- present) Questions How can artifacts operate under their own control? Contributed methods and results knowledge representation grammarThe Foundations of AIComputer Engineering(1940-present) Questions How does language relate to thought? Contributed methods and results b

35、uilding fast computersThe History of artifical intelligenceTimeline of major AI events The History of artifical intelligenceEvidence of AI folklore can be traced back to ancient EgyptWith the development of the electronic computer in 1941, the technology finally became available to create machine in

36、telligence. The term artificial intelligence was first coined in 1956, at the Dartmouth conferencesince then Artificial Intelligence has expanded because of the theories and principles developed by its dedicated researchers. Through its short modern history, advancement in the fields of AI have been

37、 slower than first estimated. The History of artifical intelligenceProgress continues to be made. From its birth 6 decades ago, there have been a variety of AI programsThey have impacted other technological advancements. The History of artifical intelligenceHistory list PrecursorsThe gestation of AI

38、(1943-1955)The birth of AI(1956)Early enthusiam,great expectation(1952-1969)A dose of reality(1966-1973)Knowledge-based systems:The key to power?(1969-1979)AI becomes an industry(1980-present)The return of neural network(1986-present)AI becomes a science(1987-present)The emergence of intelligent age

39、nts(1995-present)Availability of Very Large Data Sets (2001-present)The History of artifical intelligence Precursors The 1949 innovation of the stored program computer made the job of entering a program easier Advancements in computer theory lead to computer science, and eventually Artificial intell

40、igence. The History of artifical intelligenceAl-Jazaris programmable automata AutomatonsFormal reasoningComputer scienceThe History of artifical intelligenceThe gestation of AI (1943-1955)1943 McCulloch&Pitts The first AI work: artificial neural net, proved equivalent to Turing machine 1949 Donald H

41、ebb Hebbian learning, demonstrated a simple updating rule for modifing the connection strengths between neurons-remains an influential model to this daysThe History of artifical intelligence1950 Alan TuringHis article “Computing Machinery andIntelligence” is the first complete vision of AI.In this a

42、rticle,he introduce the Turing test, machine learning, genetic algorithms and reinforcement learning.1951 Minsky&Edmonds SNARC,the first neural network computerThe History of artifical intelligenceThe birth of Artificial intelligence(1956) Dartmouth Summer Research Conference on Artificial Intellige

43、nce in 1956 symbolizes the birth of AI. The birthplace of AI: Dartmouth CollegeThe History of artifical intelligence John McCarthy coined the term “AI” Why it is AI ? Perhaps “Computational rationality” would be better. Newell & Simon presented LOGIC THEORIST (LT) program It is a reasoning program,

44、capable of thinking non- numerically. It could prove most of theorems like Russell and Whiteheads Principia Mathematica.The History of artifical intelligence Why AI become a separate field? 1. AI from the start embraced the idea of duplicating human faculties, self-improvement and language use. None

45、 of the other fields were addressing these issues. 2. AI is the only one of these fields that is clearly a branch of computer science and AI is the only field to attempt to build machines that will function autonomously in complex & changing environments. The History of artifical intelligenceJohn Mc

46、Carthy In 1956, John McCarthy regarded as the father of AI, organized a conference to draw the talent and expertise of others interested in machine intelligence for a month of brainstorming. He invited them to Vermont for The Dartmouth summer research project on artificial intelligence. From that po

47、int on, because of McCarthy, the field would be known as AI.Although not a huge success, the Dartmouth conference did bring together the founders in AI, and served to lay the groundwork for the future of AI research. The father of AI: John McCarthy The History of artifical intelligenceEarly enthusia

48、sm, great expectations(1952-1969) 1957 Herb SimonIt is not my aim to surprise or shock you but the simplest way I can summarize is to say that there are now in the world machines that think, that learn and that create. Moreover their ability to do these things is going to increase rapidly untilin th

49、e visible futurethe range of problems that can handle will be coextensive with the range to which human mind has been applied.The History of artifical intelligence1958 John McCarthy defined the high-level language Lisp, which is the second-oldest major high-level language in cerrent use.1965 : J.A.

50、Robinson invents the resolution principle, basis for automated theorem provingA machine can never do XPart of the list of X include: fall in love, enjoy food, diversified behavior like humanThe History of artifical intelligenceIntelligent reasoning in Microworlds (such as Blocks world)Marvin Minsky

51、pointed out that: successful sciences were often best understood using simplified models like frictionless planes or perfectly rigid bodies. Much of the research focused on the so-called blocks world, which consists of colored blocks of various shapes and sizes arrayed on a flat surface .The History

52、 of artifical intelligence The Blocks worldThe History of artifical intelligenceA dose of reality(1966-1973)1965 : Weizenbaums ELIZADifficulties in automated translation, “the spirit is willing but the flesh is weak” - “the vodka is good but the meat is rotten”Limitations of Perceptrons discovered (

53、XOR problem)Machine evolution (now Genetic Algorithms)Systems for microworlds dont scale up for real applications XOR problemSeparate and Perceptron separates samples in two class by a straight line (y=AX)No single straight line can do itPerceptron fails this simple XOR problemThe History of artific

54、al intelligence1973 Lighthill report -cut AI fundingEarly AI systems turned out to fail miserably when tried out on wider selection of problem and on more difficult problems:most early programs contained little or no knowledge of their subject matter, they succeeded by means of simple syntactic mani

55、pulations. the intractability of many of the problem that AI was attempting to solvesome fundamental limitation on the basic structures being used to generate intelligence behaviorThe History of artifical intelligenceKnowledge-based systems: The key to power? (1969-1979)weak methodsA geneal-purpose

56、search mechanism trying to string together elementary reasoning steps to find complete solutions.Why need Knowledge?Weak methods do not scale up to large or difficult problem instances.The alternative to weak methods is to use more powerful domain-specific knowledgeallows larger reasoning stepsmore

57、easily handle typically occuring cases in narrow areas of expertise.The History of artifical intelligenceDendral The first successful knowledge-intensive system. Inferring molecular structure from the information provided by a mass spectrometerThe Heuristic Programming Project(Hpp) to investigate th

58、e extent to which the new methodology of expert systems could be applied to other areas of human expertise.MYCIN to diagnose blood infections.With about 450 rules.MYCIN was able to perform as well as some experts, and considerably better than junior doctorsThe History of artifical intelligenceWinogr

59、ads SHRDLU system Knowledge used in the area of understanding natural language. It was able to overcome ambiguity and understand pronoun references, but mainly because it was designed specifically for one areathe blocks world. Many programs would understanding natural language. But all on specific a

60、rea,such as representing stereotypical situations,describing human memory organizationThe History of artifical intelligenceThe widespread growth of application to realworld caused a concurrent increase in the demands for workable knowledge representation schemes.A large number of different represent