迈向第三代人工智能

被引:173
|
作者
张钹 [1 ]
朱军 [1 ]
苏航 [1 ]
机构
[1] 清华大学人工智能研究院
关键词
人工智能; 符号主义; 连接主义; 双空间模型; 单空间模型; 三空间模型;
D O I
暂无
中图分类号
TP18 [人工智能理论];
学科分类号
081104 ; 0812 ; 0835 ; 1405 ;
摘要
人工智能(artificial intelligence, AI)自1956年诞生以来,在60多年的发展历史中,一直存在两个相互竞争的范式,即符号主义与连接主义(或称亚符号主义).二者虽然同时起步,但符号主义到20世纪80年代之前一直主导着AI的发展,而连接主义从20世纪90年代才逐步发展起来,到21世纪初进入高潮,大有替代符号主义之势.今天看来,这两种范式只是从不同的侧面模拟人类的心智(或大脑),具有各自的片面性,依靠单个范式不可能触及人类真正的智能.需要建立新的可解释和鲁棒的AI理论与方法,发展安全、可信、可靠和可扩展的AI技术.为实现这个目标,需要将这两种范式结合起来,这是发展AI的必经之路.本文将阐述这一思想,为叙述方便,我们称符号主义为第一代AI,称连接主义为第二代AI,将要发展的AI称为第三代AI.
引用
收藏
页码:1281 / 1302
页数:22
相关论文
共 13 条
  • [1] Big Learning with Bayesian methods[J]. Jun Zhu,Jianfei Chen,Wenbo Hu,Bo Zhang.National Science Review. 2017(04)
  • [2] Building Watson: An Overview of the DeepQA Project[J] . Ferrucci,David,Brown,Eric,Chu-Carroll,Jennifer,Fan,James,Gondek,David,Kalyanpur,Aditya A,Lally,Adam,Murdock,J William,Nyberg,Eric,Prager,John,Schlaefer,Nico,Welty,Chris.AI Magazine . 2010 (3)
  • [3] A fast learning algorithm for deep belief nets
    Hinton, Geoffrey E.
    Osindero, Simon
    Teh, Yee-Whye
    [J]. NEURAL COMPUTATION, 2006, 18 (07) : 1527 - 1554
  • [4] Long short-term memory
    Hochreiter, S
    Schmidhuber, J
    [J]. NEURAL COMPUTATION, 1997, 9 (08) : 1735 - 1780
  • [5] Methodological solipsism considered as a research strategy in cognitive psychology 1[J] . J. A. Fodor.Behavioral and Brain Sciences . 1980 (1)
  • [6] Neocognitron: A self-organizing neural network model for a mechanism of pattern recognition unaffected by shift in position[J] . Kunihiko Fukushima.Biological Cybernetics . 1980 (4)
  • [7] Maximum Likelihood from Incomplete Data via the EM Algorithm[J] . A. P. Dempster,N. M. Laird,D. B. Rubin.Journal of the Royal Statistical Society. Series . 1977 (1)
  • [8] Computer science as empirical inquiry[J] . Allen Newell,Herbert A. Simon.Communications of the ACM . 1976 (3)
  • [9] A logical calculus of the ideas immanent in nervous activity[J] . Warren S. McCulloch,Walter Pitts.The Bulletin of Mathematical Biophysics . 1943 (4)
  • [10] Posterior sampling for multi-agent reinforcement learning solving extensive games with imperfect information .2 Zhou Y C,Li J L,Zhu J. Proceedings of the International Conference on Learning Representations (ICLR) . 2020