Intelligence, - Stanford University

2y ago
21 Views
2 Downloads
1.19 MB
16 Pages
Last View : 19d ago
Last Download : 2m ago
Upload by : Troy Oden
Transcription

07 36 MON 25 APR 77"Page 1 AVRON AIH.AI HANDBOOK OUTLINEINTRODUCTIONA.Intended AudienceThis handbook is intended for two kinds ofcomputer science students interestedin learning more about artificial intelligence, andengineers in search of techniques and ideas thatmight prove useful in applications programs.audience;B.Suggested Style For ArticlesThe following is a brief checklist that may providesome guidance in writing articles for thehandbook. It is, of course, only a suggested list.Start with 1-2 paragraphs on the central idea orconcept of the article. Answer the question "whatis the key idea?"ii) Give a brief history of the invention of theidea, and its use in A. I.iii) Give a more detailed technical descriptionof the idea, its implementations in the past,and the results of any experiments with it. Tryto answer the question "How to do it?.iv) Make tentative conclusions about the utility andlimitations of the idea if appropriate.v) Give a list of suitable references.vi) Give a small set of pointers to related concepts(general/overview articles, specific applications, etcvii) When referring in the text of an article to a termwhich is the subject of another handbook article,surround the term by 's; e. g. Production Systems-*-.i)"C.Coding Used In This OutlineThis outline contains a list of themajorareas of artificial intelligence covered in thehandbook. At the lowest level, the outline showsarticle titles either contained or needed. Inthe case of an article that is needed* the notationNEED #3 follows the proposed focus of the")

07:36 MON 25 APR 77Page 2the article, where # is a number in the intervalLow numbers indicate little expecteddifficulty with the article, whereas high numbersindicate a potentially difficult article. Forexample, an article on a specific system, where onlya minimal amount of reading is required wouldrate approximately 4, whereas an overview articlewould likely rate 8 or greater. In the case ofarticles which already exist in the handbook, thenotation done #3 is used, where low numbers indicatethat the article needs only minorandhigh numbers indicate that major modifications arerequired. For example, repair of typographicalerrors and wording could be expected to rate 0-2.Correction of errors in the article might rate 3—6,and major rewrites which require considerable readingwould likely rate 7-10.It should be noted that the real difficulty involvedwritinginan article is highly dependent on the a prioriknowledge of its author."*****This has been a preamble. Now for some areas not coveredin the other subject outlines*****A General View of Artificial" AVRON AIH.IntelligencePhilosophyNEED ERLS3This article might address the kinds of questionsraised by Turing's article (CAT), Dreyfus'sbook, the rebuttals, Lighthill's critique,McCarthy's reply, and so on.Relationship to SocietyThis might touch on sciencemisconceptions, the Delphi survey,NEED EPMc3popularand so on.HistoryNEED EPMc3Perhaps start with Cybernetics, the Dartmouthand so on. See HPS appendix. Alsonote the major centers, their focus andpersonalities. Note the role of ARPA funding on theresearch, the ties to DEC machines and so on.Conferences and PublicationsNEED 53AI journal,MI books, IJCAIproceedings,Cognitive Psychology, someSMC), ComputationallEEE (Computers,Linguistics, Special interest conferences: robotics,cybernetics, natural language.Note the tech note unofficial type documents"3

07 36 MON 25 APR 77"IIPage 3 AVRON AIH.HEURISTIC SEARCHA.Heuristic Search OverviewNEED ENillAlgorithmic presentation of "heuristic search" procedure.Heuristics for choosing promising nodes to expand next*heuristics for choosing operators to use to expand a nodeMeta-rules: using heuristics to choose relevant heuristics.Pervasive character of the combinatorial explosion.Arguments (both formal and intuitive) supporting the use ofheuristic search to muffle this explosion.Formal: Completeness ofKnuth 's recent work onalpha-beta search.Opportunities for future researchWhere do heuristics come from?(see Simon's current work; meta-rules; meta-meta-.? experiencesheuristicsbasedonModifying(see Berliner's current work)Working with symbolic, rather than numerical, values for nodesCoding heuristics as production rules(e.g.: view Mycin as a heuristic search)Situations NOT suited to attack by heuristic searchTypically: non-exponential growth process; no search anyway(e.g.* finding roots of a quadratic equation)Identity problemsDisguising Heuristic Search as something elseDisguising something else to appear to be a Heuristic Search."B.Search Spaces1. Overviewdone E 3The concept of a search space; how a search spacecan be used to solve (some) problems; differentrepresentations, different spaces2.State-space representationdone E63here,oughtwhichtobeunified3E2 articles exist3.Problem-reduction representation4. AND-OR trees and graphsC."Blind" Search Strategies1. Overview"2.Breadth-first searching3.Depth-first searching4.Bi-directional searchingdiscuss heuristics.5.Minimaxing6.Alpha-Beta searchingMI articles by ira Pohldone E33

Page 407: 36 MON 25 APR 77D.Using AVRON AIHHeuristics to Improve the Searchdone 731. OverviewofThe ideaa heuristicThe idea of a heuristic evaluation functionsavings in change of representation.2.done 43Best-first searching(Ordered-search ) but need to add: Martelli'swork (ask Nils for a draft of this)speech rec: IJCAI-3 (Paxton), Reddy 's book3.Hill climbingdone 33done E334. Means-ends analysis5. Hierarchical search, planning in abstract spaces done E 3Abstrips (Sacerdoti)6. Branch and bound searchingdone 437. Band-width searchingHarrisAI journaldone 3-E.Programs employing(basedon)heuristic search1. OverviewNEED 73limitations.Comparison of systems.Results &(This first article should be written later as anintroduction to the following articles. )2.Historically important problem solversa)done 3GPSdone Eb ) Strips,c)Gelernter's Geom.Program3NEED 333

07 37 MON 25 APR 77111Page 5 AVRON AIH.Natural LanguageA.Overview1.early machine translationFailures of straight forward approachesdon e E5 3History and Development of N. L.done E 3Main ideas (parsing, representation)comparison of different techniques.mention ELIZA, PARRY.Include Baseball, SadSIR and Student articles here,see Winograd 's Five Lectures, Simmon 's CACM articles.2.B. Representation of Meaning(seeC.2.3.4.5.6.7.8.D.—HIP)Syntax and Parsing Techniques1."section VIIoverviewsa. formal grammarsb. parsing techniquesaugmented transition nets, WoodsShrdlu's parser (systemic grammars)Case Grammars Bruce (AI Journal, 1/76)CHARTSwell formed substringssyntaxGSP& parserproblem understandingH. Simontransformational us Natural Language systems1.Winograd2. SCHOLAR3. SOPHIEE.F."E33C63E33E53E 3E 3E 3E 3E53NEED E53done E 3done E 3Current translation techniquesWilks' work, commercial systemsText Generating systemsSheldon Kleimdone E 3(VauquoisSimmons and Sloan)(indone E 3S&C )

Page 607:37 MON 25 APR 77 AVRON AIH3IV. AI LanguagesA.Earlylist-processing languagesdone 33overview articleIPL, SLIP, SNOBOL, FLPLlanguages likerecursion,list structure,ideas:associative retrievalB. Language/system features0.Overview of current list-processinglanguages1.Control structures, what languages theyare in and examples of their use.done E 3done E3Bactracking (parallel processing)Demons (pseudo-interrupts )Pattern directed computation2. Data done E 3lists,.)Once again, examples of their useis important here.3.Pattern Matching in AI languagessee Bobrow & Raphael4. Deductive mechanismssee Bobrow & RaphaelC.NEED 63NEED E53Current languages/systems1.2.3.4.5.6.7.8.the basic ideaINTERLISPQLISP nedonedone 23E 3 33 23 23 23 3 3CONNIVERSLIPPOP-2(perhaps do whole POP-10 system, includingPQPLER. See AlcockSLondon. Ontario)done 39. SNOBOL(see thm.10. QA3/PROLOGUEprov.)

Page 737 MON 25 APR 77 kV. AVRQN AIH OUTLINE; 3AUTOMATIC PROGRAMMINGA.OverviewB.Program Specification Techniquesi.e. how does the user describe the program tobe synthesized?an overview article including various methods done E 3see SAFE system (ISI), Green's tech. report,Smith 's graphic specif ication, andinclude general remarks on the high-levellanguage methodsdone 73—C.Program Synthesis techniquesdone 3given a description of the program in somegenerate the actual program OVERVIEW whichinductive methods, problem solving methods, andknowledge-based programs. Places synthesis inperspective of AP process.New info on Simon'sHeuristic Compiler, and Manna & Waldinger'swork on plan modification.-—1. Traces2. Examplesdone 33done 33at U. of"(include3. Problem solving applications toa. Sussman 's HackerBiggerstaffb.Washington)APProgram Synthesis by Theorem ProvingGreen and Waldinger (PROW)—done 23done 334.Codification of Programming KnowledgeNEEDE63see C.Green's work, Darlington, Rich & Shrobe5.Integrated AP SystemsNEEDE73see Lenat's original work, Heidorn, Martin'sPSI at SAILD.Program optimization techniquesdone E 3How to turn a rough draft into an efficientprogram.See Darlington & Burstall, Low, Wegbreit, Kant.E.Programmer's aids(Interlisp's DWIM,F.Program verification(IJCAI 3)done E3etc)done 3

37 MON 25 APR 77"VI.Page 8 AVRON AIHTHEOREM PROVINGA.OverviewB.Resolution Theorem Proving1. Basic resolution methoddone 432.Syntactic ordering strategiesdone 233.Semantic & syntactic refinementdone 23 C4.C.done E 3other strategies?}Non-resolution theorem proving1. Natural deduction2.done 33Boyer-Moore3. LCFD.Uses of theorem proving1.Use in question answering2. Use in problem solvingdone E33done 633.Theorem Proving languages(QA3, Prologue)done E 34.Man-machine theorem provingdone 3(Bledsoe)E.Predicate CalculusF.Proof checkersdone 533

07:37 MON 25 APR 77VII.Page 9Human Information Processing(seeA.— AVRON AIH.PsychologyPerry's outline for details andreferences)PerceptionNEED EPWT3An overview of relevant work in psychologyon attention, visual and auditory perception,pattern recognition.Applied perception (PERCEIVER).Difficulties resulting from inability to introspect.B.Memory and Learning1. Basic structures and processes in IPPdone 3Short- and Long-term memory, Rehearsal, Chunking,Recognition, Retrieval, recall, Inference andquestion-answering, Semantic vs. episodic memory.Interference and forgetting, Type vs. token nodesSimonSciences of the Artificial-2. Overview of memory models, RepresentationSpHow to get to the airport:the various models.a.NEED EAB3A comparison ofAssociative memory models1.semantic netsdoneNash-Weber (BBN)Shapiro, Hendricks (SRI), Wood's articlein Bobrow &Simmons (S&C)HAM (Anderson & al analysisdoneSchank (conceptualdependency), (MARGIE), G. MillerEPAMNEEDdoneQuery languages(IBMSJ)(1968),TedWood'sCoddE 3Quillian (TLC),2.3.4.5.6.b. 3 3E 3EEAF3 3Other representations1. Production systems2. Frame systems (Minsky, Winograd)3. Augmented Transition Networks4. Scripts (Schank, Abelson)done 13done 73done 33done 33

07:38 MON 25 APR 77"C. AVRON AIH.Page 10NEED EPWT3Psychol inguisticsA prose glossary including:Competence vs. performance models, Phonology vs.syntax vs. semantics vs. pragmatics, Surface vs.deep structure, Taxonomic grammars, generative grammars,transformational grammars, Phrase-structure rules,lexical entriestransformation rules.Parsing vs. generation, Context-free vs. Context-sensitivegrammars, Case systems (e. g. , Bruce AI article)D.Human Problem Solving—done E 3Overviewdone 13NEED E631. PBG's2. Human chess problem solvingE.Behavioral Modeling1. Belief SystemsAbelson* McDermott2. Conversational Postulates3. Parry""done E(Grice,TW)3NEED E53NEED E533

07: 38 MON 25 APR 77VIIIPage 11 AVRON AIHVISIONNEED EPH3OverviewThis article should discuss the early work in vision;its roots in pattern recognition, character recognition,Pandemonium, Perceptrons and so on. (ie. the pre-Robertswork). It should discuss the main ideas of modern visionwork as a leadin to the more specific articles, forexample the use of hypothesis, model, or expectationdriven strategies. It should also discuss the way inwhich the focus of the field flip-flops from front endconsiderations to higher level considerations withtime.Polyhedral or Blocks World VisionAn overview article should include the majorideas in this work together with briefsummaries of the work of the major investigatorsIn addition, separate articles should be writtenon the work of those listed below."done E 3Overview(Roberts,Huffman andShirai and others listedbelow)Guzmandone E23Falkdone E 3done E 3WaltzgeneralarticleshouldcontainmoreThismaterial on constraintdrawnpossibly from Montenari and FikesThis exhausts my list. Please add others or delete someof mine if appropriate.It has been suggested EBolles3 that the most instructivemethod of writing these articles would be to providesimple examples of the problems attacked by the variousprograms.Scene Analysis"NEED iledstrategies used, and the present state of theart.The following articles should be written or modified todescribe the specialized tools of scene analysis.See Duda and Hart.3

07:38 MON 25 APR 77Template Matching(a non-mathematical"Page 12 AVRON AIHdone E 3description)done E43Edge Detectiondone E73Homogeneous CoordinatesmodifiedtoincludeThis article should bethe general questions of the perspectivecamera calibration, and so on.Line Descriptiondone 43Noise Removaldone 43Shape Descriptiondone 43Region Growing (Yakamovsky,ky,Olander)done 33Contour FollowingNEED 43Spatial FilteringNEED 43Front End Particularsdone 3This article should contain some description of thecompressimethods and effects of compressionand quantizationexample.for"D.E.Syntactic Methodsdone E 3Descriptive MethodsSee Duda and Hart, andnd Winstondone E 3on SystemsRobot and Industrial VisionOverview and State of thee Artdone 3HardwareNEED EB3Pattern RecognitionIt's not clear just wheree this discudiscussion should go, orrequired.detailiswhat level ofdone EB3OverviewuprefocussedandcleanedThis article needs to beF."Statistical Methods and ApplicationsNEED E93Descriptive Methods and ApplicationsNEED EB3MiscellaneousMultisensory ImagesNEED E73Perceptronsdone E33

144fcIX.38 MON 25 APR 77Page 13 AVRON AIHSPEECH UNDERSTANDING SYSTEMSOverview(includea mention of ac.proc.)done E33Integration of Multiple Sourcesof KnowledgeNEED 93For example the blackboard of the HEARSAY II systemHEARSAY Idone 43HEARSAY IIdone 53SPEECHLISdone 23SDC-SRI System(VDMS)done 3DRAGONdone 63Jim Baker's original system plus Speedy-Dragon byBruce Lowerre. This article is a little harder thanthe other system articles because the methods usedmay be unfamiliar to some.3

38 MON 25 APR 77"*Page 14 AVRON AIHROBOTICSdone E 3OverviewThis article should discuss the central issues andits history, and thedifficulties of thepresent state of the art.done E 3Robot Planning and Problem SolvingFor example, STRIPS and ABSTRIPS. This articlecould be quite general depending on the point ofview taken.Armsdone EExplain the difficulties of control at the bottomlevel, system integration, obstacle avoidanceand so on. Also note the problems with integrationof multi-sensory data, for example vision andtouch feedback.Day Industrial Robotsdone 3Robotics Programming LanguagesFor example WAVE, and AL(a short article)done E 3Present""33

38 MON 25 APR 77XIPage 15 AVRON AIH. OUTLINE; 3Applications-oriented AI researchAn overview article.What are the attributesNEED 83of a suitable domain? Custom craftingtheory vs. actual use. (See EAF: 225 notes, 1972)-A.Chemistry1.Mass spectrometry2.Organic Synthesis(DENDRAL,meta-dendral )done 63OverviewNEHD 83Summarize work of Wipke, Corey, Gelernter, and SridharanB.Medicine1.2.C."D.doneEl 3Summarize DIALOG(PopIe ), CASNET(Kuli kowski ), NEEDE73Pauker's MIT work, and the Genetics counsellingprogramsPsychology and PsychiatryProtocol Analysis (Waterman andNewell)REDUCEMACSYMA(mentionSAINT)Assemblyline balancing (Tonge)Electric power distribution systems(MI)done 3done E 3done E 3done 3done 3done 3MiscellaneousOverview of music composition and aesthetics"done 3done 3Miscellaneous1. LUNAR2. EducationPapert, or more ?3. SRI computer-based consultation4. RAND—RITA production rule system forintelligent interface softwareI.3Business and Management Science Applications1.2.F.done EMath systems1.2.E.MYCINdone 73

07:39 MON 25 APR 77Page 16 AVRON AIHReasoning by analogydone 43Intelligence augmentationdone 53Chessdone 53XIII. Learning and Inductive InferenceOverviewNEED 93Samuel Checker programdone 3Winstondone 23Pattern extrapolation problemsOverview of InductionAQVAL (Michalskiat U.Ill)—done E 33

Main ideas (parsing, representation) comparison of different techniques. mention ELIZA, PARRY. Include Baseball, Sad Sam, SIR and Student articles here, see Winograd's Five Lectures, Simmon's CACM articles. B. Representation of Meaning (see section VII — HIP) C. Syntax and Parsing Techniques 1. overviews a. formal grammars b. parsing .

Related Documents:

SEISMIC: A Self-Exciting Point Process Model for Predicting Tweet Popularity Qingyuan Zhao Stanford University qyzhao@stanford.edu Murat A. Erdogdu Stanford University erdogdu@stanford.edu Hera Y. He Stanford University yhe1@stanford.edu Anand Rajaraman Stanford University anand@cs.stanford.edu Jure Leskovec Stanford University jure@cs.stanford .

Computer Science Stanford University ymaniyar@stanford.edu Madhu Karra Computer Science Stanford University mkarra@stanford.edu Arvind Subramanian Computer Science Stanford University arvindvs@stanford.edu 1 Problem Description Most existing COVID-19 tests use nasal swabs and a polymerase chain reaction to detect the virus in a sample. We aim to

Domain Adversarial Training for QA Systems Stanford CS224N Default Project Mentor: Gita Krishna Danny Schwartz Brynne Hurst Grace Wang Stanford University Stanford University Stanford University deschwa2@stanford.edu brynnemh@stanford.edu gracenol@stanford.edu Abstract In this project, we exa

Stanford University Stanford, CA 94305 bowang@stanford.edu Min Liu Department of Statistics Stanford University Stanford, CA 94305 liumin@stanford.edu Abstract Sentiment analysis is an important task in natural language understanding and has a wide range of real-world applications. The typical sentiment analysis focus on

ARTIFICIAL INTELLIGENCE LAB STANFORD UNIVERSITY SCHOOL OF ENGINEERING / COMPUTER SCIENCE DEPARTMENT 1 Stanford Artificial Intelligence Lab FEBRUA RY 2 019. 2 STANFORD UNIVERSITY Maneesh Agrawala Professor, Computer Science; Director, the Brown Institute for Media Innovation Computer graphics, human-computer interaction, visualization To improve the effectiveness of media of all kinds Gill .

Stanford Health Care Organizational Overview 3 Contract Administration is a Shared Service of Stanford Health Care to Eight Other Stanford Medicine Entities Stanford Health are ("SH")is the flagship academic medical center associated with the Stanford University School of Medicine. SHC has 15,232 employees and volunteers, 613 licensed

Mar 16, 2021 · undergraduate and graduate students, faculty, staff, and members of the community. Anyone interested in auditioning for the Stanford Philharmonia, Stanford Symphony Orchestra, or Stanford Summer Symphony should contact Orchestra Administrator Adriana Ramírez Mirabal at orchestra@stanford.edu. For further information, visit orchestra.stanford.edu.

STANFORD INTERNATIONAL nANK, LTD., § STANFORD GROUP COMPANY, § STANFORD CAPITAL MANAGEMENT, LLC, § R. ALLEN STANFORD, JAMES . M. DAVIS, . The false data has helped SGC grow the SAS program from less than 10 million in around 2004 to . I : over 1.2 billion, generating fees for SGC (and ultimately Stanford) in excess of 25 million. .