1、浅谈人工智能的下个十年1人工智能的第三次浪潮2人人工工智智能能历历史史3人工人工智能智能领领域发展域发展趋趋势势Powered byClaude ShannonShannon,Claude E.XXII.Programming a computer for playing chess.Philosophical magazine 41.314(1950):256-275.1950计算机象棋博弈1954图灵测试Alan TuringTuring,Alan M.Solvable and unsolvable problems.Science News-ens.fr 39 (1954).4人工人工智
2、能智能领领域发展域发展趋趋势势Powered by1956达特茅斯会议McCarthy,J.,et al.Dartmouth Conference.Dartmouth Summer Research Conference on Artificial Intelligence.19561959一般问题解决器JohnMarvin McCarthy MinskyNathan RochesterClaudeShannonHerbertJ.C.ShawAllen NewellSimonNewell,A.;Shaw,J.C.;Simon,H.A.(1959).Report on a general pro
3、blem-solving program.Proceedings of the International Conference on Information Processing.pp.256264.5人工人工智能智能领领域发展域发展趋趋势势Powered byDaniel BobrowBobrow,Daniel G.Natural language input for a computer problem solving system.(1964)1964 理解自然语言输入1966 ELIZA人机对话Joseph Weizenbaum Weizenbaum,Joseph.ELIZAa co
4、mputerprogram for the study of natural language communication between man and machine.Communications of the ACM 9.1(1966):36-45.6人工人工智能智能领领域发展域发展趋趋势势owered by1968 世界首个专家系 统 DENDRALEdward FeigenbaumBuchanan,Bruce,Georgia Sutherland,and Edward A.Feigenbaum.Heuristic DENDRAL:a program for generating ex
5、planatory hypotheses in organic chemistry.Defense Technical Information Center,1968.7人工人工智能智能领领域发展域发展趋趋势势Powered byRandall DavisDrew McDermott,Jon DoyleMcDermott D,Doyle J.Non-monotonic logic IJ.Artificial intelligence,1980,13(1):41-72.1976 大规模知识库构建与维护Applications of meta level knowledge to the cons
6、truction,maintenance and use of large knowledge basesM.Stanford University,Computer Science Department,AI Laboratory,1976.1980 非单调逻辑8人工人工智能智能领领域发展域发展趋趋势势Powered by1980 计算机战胜双陆棋世界冠军Hans BerlinerBerliner H J.Backgammon computer program beats world championJ.Artificial Intelligence,1980,14(2):205-220.9
7、人工人工智能智能领领域发展域发展趋趋势势owered by1987 基于行为的机器P人学Rodney BrooksBrooks R.A robust layered control system for a mobile robotJ.Robotics and Automation,IEEEJournal of,1986,2(1):14-2310人工人工智能智能领领域发展域发展趋趋势势owered by1987 自我学习双陆棋P程序Gerry TesauroTesauro G.TD-Gammon,a self-teaching backgammon program,achieves maste
8、r-level playJ.Neural computation,1994,6(2):215-219.11人工人工智能智能领领域发展域发展趋趋势势Powered byTim Berners-LeeBerners-Lee,Tim.Semantic web road map.(1998).McGuinness,Deborah L.,and Frank Van Harmelen.OWL web ontology language overview.W3C recommendation 10.2004-03(2004):10.1998 语义互联网路线图2004 OWL语言12人工人工智能智能领领域发展
9、域发展趋趋势势Powered byGeoffrey HintonLe,Quoc V.,et al.Building high-level features using large scale unsupervised learning.arXiv preprint arXiv:1112.6209(2011).2006 深度学习Hinton,Geoffrey E.,Simon Osindero,and Yee-Whye Teh.A fast learning algorithm fordeep belief nets.Neural computation 18.7(2006):1527-1554
10、.2011 高层抽象特征构建13人工人工智能智能领领域发展域发展趋趋势势Powered by2009 谷歌自动驾驶汽车Sebastian ThrunMarkoff,John.Google cars drive themselves,in traffic.The New York Times 10(2010):A1.14人工人工智能智能领领域发展域发展趋趋势势Powered by2011 沃森获得Jeopardy冠军IBMs WatsonMarkoff,John.Computer program to take on Jeopardy!.The New York Times (2009).201
11、1 自然语言问答Apples Siri Sadun,Erica,and Steve Sande.Talking to Siri:Learning the Language of Apples Intelligent Assistant.Que Publishing,2013.15人工人工智能智能领领域发展域发展趋趋势势Powered by16人人工智能工智能近近10 年年17AI趋势：从感知到认知 From perceptron to cognitionComputingPerceptionCognitionStorage&ComputingRecognize text,images,obje
12、cts,voicesOrganize and generate knowledge,reasoning18Artificial IntelligenceAlphaGoImage recognitionSelf-driving1920DDPG(2015)A3C(2016)Perceptron(1958)Frank Rosenblatt Cornell University psychologistBPNN/MLP(1986)Hopfield Network(1982)recurrent&feedbackGeoffery Hinton University of Toronto deep lear
13、ningNeocognitron(1980)convolution&poolingLeNet/CNN(1998)Yann LecunNew York University deep learningAlexNet(2012)Relu,dropout&biggerVGG(2014)GoogLeNet(2015)ResNet(2016)Kaiming He MSRA=FAIRcomputer visionDenseNet(2017)RBM(1986/2006)stackDeep Belief Nets(2006)VAE(2013)AutoEncoder(1989/2006)Denosing Aut
14、oencoder(2008)Variational InferenceMax WellingUniversity of Amsterdam statistical learningGAN(2014)DCGAN(2014)WGAN(2017)PGGAN(2017)Ian Goodfellow Google Braindeep adversarial learningRNN/LSTM(1997)Jrgen SchmidhuberIDSIAUniversal AISeq2Seq(2014)RNN in Speech Recognition(2013)Yoshua Bengio University
15、of Montreal Deep learningNeural Probabilistic Language Model(2003)word2Vec(2013)SeqGAN(2017)LeakGAN(2018)Character CNN(2015)self-attention(2017)Deep Q-learning(2013)AlphaGo(2016)Double DQN(2015)Dueling Net(2016)David Silver DeepMindReinforcement learningAlpha Zero(2017)Capsule Nets(2017)算法BERT Pre-t
16、rain Fine tune Beat all state-of-the-arts on 11 NLP tasks in 201821XLNet Autoregressive Model Beat BERT in 201922ALBERT A Lite BERT Parameter-reduction techniques Beat XLNet and all the others23Video-to-Video Synthesis The best video synthesis performance24graph_net By DeepMind25MoCo Unsupervised vi
17、sual representation learning Momentum contrastive learning Outperform its supervised pre-training counterparts2627SimCLR Simplified contrastive learning framework Outperform previous self-supervised and semi-supervised methods on ImageNet人人工工智智能能未未来来28第三代人工智能的理论体系 早在2015年，张钹老师就提出第三代人工智能体系的雏形；2017年DA
18、RPA发起XAI项目，从可解释的机器学习系统、人机 交互技术以及可解释的心理学理论三个方面，全面开展可解释性 AI系统的研究 2018年底，正式公开提出第三代人工第三代人工智智能能的的理理论论框框架架体体系系 建立可解释、鲁棒性可解释、鲁棒性的人工智能理论和方法 发展安全、可靠、可安全、可靠、可信信及可扩及可扩展展的人工智能技术 推动人工智能创新创新应用 具体实施路线图 与脑科脑科学学融合，发展脑启发的人工智能理论 数据与知识融数据与知识融合合的人工智能理论与方法 第三代人工智能的理念在国内外 获得广泛影响力29认知图谱(Cognitive Graph)知识图谱,认知推理,逻辑表达30知识图谱
19、“Knowledge graph”由Google于2012年提出 知识工程，专家系统 CYC:世界上历史最长的AI项目(1985)Tim Berners Lee Father of WWW Turing AwardEdward Feigenbaum Father of KB Turing Award31认知图谱：算法与认知的结合Quality CafThe Quality Cafe is a now-defunct diner in Los Angeles,California.The restaurant has appeared as a location featured in a nu
20、mber of Hollywood films,including Old School,Gone in 60 Seconds,.Los AngelesLos Angeles is the most populous city in California,the second most populous city in the United States,after New York City,and the third most populous city in North America.Old SchoolOld School is a 2003 American comedy film
21、 released by DreamWorks Pictures andThe Montecito Picture Company and directed by Todd Phillips.Todd PhillipsTodd Phillips is an American director,producer,screenwriter,and actor.He is best known for writing and directing films,including Road Trip(2000),Old School(2003),Starsky&Hutch(2004),and The H
22、angover Trilogy.Alessandro Moschitti is a professor of the CS Department of the University of Trento,Italy.He is currently a Principal Research Scientist of the Qatar Computing Research Institute(QCRI)Alessandro MoschittiTsinghua UniversityTsinghua University is a major research university in Beijin
23、g and dedicated to academic excellence and global development.Tsinghua is perenniallyranked as one of the topacademic institutions inChina,Asia,and worldwide.32算法:BIDAF,BERT,XLNet 目标：理解整个文档，而不仅仅是局部片段 但仍然缺乏在知识层面上的推理能力BiDAFBERTXLNet33挑战:可解释性34 大部分阅读理解方法都只能看做黑黑盒盒:输入:问题和文档输出:答案文本块(在文档中的起止位置)如何让用户可以验证答案的
24、对错:推理路径或者子图每个推理节点上的支撑事实用于对比的其他可能答案和推理路径认知图谱:知识表示,推理和决策35和认知科学的结合Dual Process Theory(Cognitive Science)System 1 IntuitiveSystem 2 Analytic3637Reasoning w/Cognitive Graph System 1:Knowledge expansion by association in text when reading System 2:Decision making w/all the informationSystem 1IntuitiveSys
25、tem 2Analytic38CogQA:Cognitive Graph for QA An iterative framework corresponding to dual process theory System 1 extract entities to build the cognitive graph generate semantic vectors for each node System 2 Do reasoning based on semantic vectors and graph Feed clues to System 1 to extract next-hop
26、entitiesQuestionQuality cafTodd PhillipsGone in 60 secondsOld schoolLos AngelesDominic SenaSystem 2Cognitive Graphlocation featured in a number ofSystem 1inputHollywood films,icn cl ul uedsi n g Old School,Gone in 60 Secondspredict39Cognitive Graph:DL+Dual Process Theory?System 1:implicit knowledge
27、expansion1.M.Ding,C.Zhou,Q.Chen,H.Yang,and J.Tang.Cognitive Graph for Multi-Hop Reading Comprehension at Scale.ACL19.40System 2:explicit decisionCognitive Graph:DL+Dual Process TheorySystem 1:implicit knowledge expansion1.M.Ding,C.Zhou,Q.Chen,H.Yang,and J.Tang.Cognitive Graph for Multi-Hop Reading C
28、omprehension at Scale.ACL19.41System 2:explicit decision42Cognitive Graph:Representation,Reasoning,and Decision?43认知与推理Trillion-scale common-sense knowledge graphTim Berners Lee Turing Award WinnerEdward Feigenbaum Turing Award Winner*AI=Knowledge+IntelligenceBig DataKnowledgeIntelligence44Thank you！
