人工智能详解课件.ppt

1、Modern Artificial Intelligence and Its Importance in the Future WorldZengchang Qin(Ph.D.)Intelligent Computing and Machine Learning LabSchool of Automation and Electrical EngineeringBeihang UniversityShahe Campus Oct 27 2010This is ScienceGive a big picture of modern Artificial Intelligence and unde

2、rstand why it is important in the current and the future world.We have such a direction of research in the school of ASEE.To clarify the misunderstanding of A.I.from those robot movies and science fictions.About This TalkI have been working in A.I.are for the past decade.I enjoy movies and unbounded

3、 thinking.I am always intrigued by any kinds f excellent ideas from human intelligence.Feel free to ask any questions you have in mind,no guarantee to be answered.About The SpeakerMisunderstandingArtificial Intelligence(A.I.)RoboticsJohn McCarthy(Stanford)Artificial Intelligence We fear?I,RobotThe T

4、hree Laws of Robotics by Issac Asimov are as the follows:A robot may not injure a human being or,through inaction,allow a human being to come to harm.A robot must obey any orders given to it by human beings,except where such orders would conflict with the First Law.A robot must protect its own exist

5、ence as long as such protection does not conflict with the First or Second Law.My Philosophy of Modern A.I.Artificial Intelligence is a mathematical/computing technology that will make life better.I have been interested in making machines intelligent by designing algorithms.I may not believe that on

6、e day we can recreate human brains using silicon chips,but I believe that computing will aid our brains to do missions impossible in the future.Chinese Room ParadoxModern A.I.The Engineering Approach:Machine Learning and Data MiningPattern Recognition,Computer vision and Image ProcessingDistributed

7、A.I./multi-agent systemsBiometrics and computer forensicsNatural Language ProcessingIntelligent Search and Information RetrievalComputational Cognitive ScienceComputational Neuroscience and bioinformaticsComputational Cognitive ScienceComputational/Behavior FinanceBehavior Targeting and Personal Ser

8、vicesDigital Advertisements/recommendation systemsPhilosophy of Machine LearningMachine Learning search in the hypothesis space to find the ones that match the data.Using Occams razor,we choose the simplest one.William of Ockham(or Occam)was a 14th-century English logician and Franciscan friar whos

9、name is given to the principle that when trying to choose between multiple competing theories the simplest theory is probably the best.This principle is known as Ockhams razor.ExampleExample 2Why Machine Learning is important?To fine the theory that explains the data,we usually prefer the simple one

10、s.Machine learning and scientific discovery share similarities.Karl Popper Logic ProgrammingLondon Underground ExampleFuzzy LogicMembership function(continuous)Membership FunctionsSome Intuition Professor of Fuzzy Logic Multi-agent SystemDistributed A.I.-coordinationData mining is the process of ext

11、racting patterns from data-Torture the data until they confess.Data is everywhere and in different types.Pattern Recognition and Data Mining Welcome to FairmontNET.stdtext font-family:Verdana,Arial,Helvetica,sans-serif;font-size:11px;color:#1F3D4E;.stdtext_wh font-family:Verdana,Arial,Helvetica,sans

12、-serif;font-size:11px;color:WHITE;  HTML and EmailsReturn-pathReceivedfromrelay2.EECS.Berkeley.EDU(relay2.EECS.Berkeley.EDU169.229.60.28)byimap4.CS.Berkeley.EDU(iPlanetMessagingServer5.2HotFix1.16(builtMay142003)withESMTPid;Tue,08Jun200411:40:43-0700(PDT)Receivedfromrelay3.EECS.Berkeley.ED

13、U(localhost127.0.0.1)byrelay2.EECS.Berkeley.EDU(8.12.10/8.9.3)withESMTPidi58Ieg3N000927;Tue,08Jun200411:40:43-0700(PDT)Receivedfromredbirds(dhcp-168-35.EECS.Berkeley.EDU128.32.168.35)byrelay3.EECS.Berkeley.EDU(8.12.10/8.9.3)withESMTPidi58IegFp007613;Tue,08Jun200411:40:42-0700(PDT)DateTue,08Jun200411

14、:40:42-0700FromRobertMillerSubjectRE:SLTheadcount=25In-reply-toToRandyKatzCcGlendaJ.Smith,GertLanckrietMessage-idMIME-version1.0X-MIMEOLEProducedByMicrosoftMimeOLEV6.00.2800.1409X-MailerMicrosoftOfficeOutlook,Build11.0.5510Content-typemultipart/alternative;boundary=-=_NextPart_000_0033_01C44D4D.6DD9

15、3AF0Thread-indexAcRMtQRp+R26lVFaRiuz4BfImikTRAA0wf3Qtheheadcountisnow32.-RobertMiller,AdministrativeSpecialistUniversityofCalifornia,BerkeleyElectronicsResearchLab634SodaHall#1776Berkeley,CA94720-1776Phone:510-642-6037fax:510-643-128924Medical Image,handwritten recognition25Sounds-fingerprints26Inte

16、lligent Search and Bio-identityMirco-array Data of GenesDrug DesignsComputer Human Interface EEG signalsStock IndexData Types fraud detectionSocial Network MiningMonitoring flu through twitter.Monitoring traffic through mobile calls.Entity Cube34Experimental Economics Vernon L.Smith for having estab

17、lished laboratory experiments as a tool in empirical economic analysis,especially in the study of alternative market mechanisms”From http:/nobelprize.org/Behavior Economics Irrational AgentsNotable for his work on the psychology of judgments and decision making,behavioral economics.Winning$10 or$100

18、0 with chance of 1%.Losing$10 or$1000 with chance of 1%Software Agents for TradingWhat is the capital of China?What is the population of Beijing?What is the population of the capital of China?Reasoning with Natural Language Evolutionary ComputingGenetic AlgorithmSir Richard Dawkins “The selfish Gene

19、s”Stochastic OptimizationCellular AutomatonWolfram was educated at Eton.At the age of 15,he published an article on particle physics4 and entered Oxford University at age 17.He wrote a widely cited paper on heavy quark production at age 18.2Wolfram received his Ph.D.in particle physics from the Cali

20、fornia Institute of Technology at age 205 and joined the faculty there.He became highly interested in cellular automata at age 21.2 Wolframs work in particle physics,cosmology and computer science earned him one of the first MacArthur awards.Decision TreesP(h|e)=P(e|h)P(h)/P(e)A Proof that everyone

21、can understandP(h,e)=P(h|e)P(e)P(e,h)=P(e|h)P(h)Bayesian StatisticsGraphical Model of Gaussian Distribution and Hiearachical Structure with Latent Variables Understanding SemanticsDemographics MS AdCenter LabCommercial Intentions of Given WebsiteIf you want to sell one,what is the best price?N97(Nok

22、ia Phone)Minority GameEI Farol BarMinority Game ModelApplicationIn Real worldThere are more than1 00 Irish music lovers but El Farol has only 60 seats.The show is enjoyable only when fewer than 60 people show up.Every people should decide weekly whether go to the bar to enjoy live music in the risk

23、of staying in a crowd place or stay at home.The rules are simple:a finite number of players have to choose between two sides;whoever ends up in the minority side is a winner.Simplified from market aiming to analyze complex financial marketCollective Behavior DecompositionSimulation Results(Li,Ma and

24、 Qin,2010)Ying Ma,Guanyi Li,Yingsai Dong and Zengchang Qin(2010),Minority game data mining for market predictions,for Stock Market Predictions,to appear in the Proceedings of AAMAS 2010.Guanyi Li,Ying Ma,Yingsai Dong and Zengchang Qin(2010),Behavior learning in minority games,To appear in the Procee

25、dings of CARE 2009.Zengchang Qin,Marcus Thint and Zhiheng Huang(2009),Ranking answers by hierarchical topic models,Proceedings of IEA/AIE 2009,LNCS 5579,pp.103-112,Springer.Zhiheng Huang,Marcus Thint and Zengchang Qin(2008),Question classification using head words and their hypernyms,The Proceedings of Conference on Empirical Methods on Natural Language Processing,pp.927-936,ACL.ReferencesNon-academicAcademic AIFuzzy Logic and Logic of ScienceNLP&ANNGA,ALIFE&Multi-agentWeb: or Google“Zengchang Qin”for my LinkedIn Profiles.Contact InformationThank you very much!Any questions?