History

 

 

 

 

 

 

 

 

 

Historial development according [Jackson 86]

1950-1965 "clásical" period

  • Gestation [McColluck y Pitts, Shannon, Turing]
  • Begin - meeting in Darmouth College in 1956 [ Minsky, McCarthy]
  • Neural Network, robotics (Shakey)
  • Searching in a space of states, heuristic, LISP
  • General Problem Solver (GPS) [Newell, Simon]
  • Games, demostration of theorems
  • Limitations of pure search, combinatory explosion.
1965-1975 "romántic'' period
  • General representation of knowledge.
  • Semantics Networks [Quillian]
  • Prototips (frames) [Minsky]
  • Perceptron [Minsky and Papert]
  • Logical [Kowalski]
  • World of blocks (SHDRLU) [Winograd]
  • knowledge of language, vision, robotics.
  • Representation difficulties "general'', problems of "toy".

1975-Current "modern'' period, "specific" Intelligence vs "general''.

  • Explicit representation of knowledge in a dominion.
  • Expert Systems or based in a knowledge.
  • Return to Neural Networks [Hopfield, Rumelhart, Hinton], genetics algoritms [Holland, Goldberg] Voice recognition (HMM), uncertainty (RB, fuzzy logic), planning, learning of real applications (medicine, finances, engineering, exploration, etc.).
  • Limitations: supercial knowledge


    According [Russell, Norvig 95]:

    Gestation (1943-1956):
    McCullock and Pitts (43), Hebb (49), Shannon
  • (50), Turing (53), Minsky and Edmonds (51). Darmouth College (56) McCarthy, Newell y Simon "The Logic Theorist''
  • Enthusiasm and great expectations (1952-1969):
  • Samuel - checkers (52), McCarthy (58):
  • Lisp, time sharing, Programs with common sense. Minsky y McCarthy in MIT. In 63 McCarthy was gone to Stanford SRI, Shakey, etc. Minsky, Evans, Student, Waltz, Winston, Winograd, etc. Works in RN: Hebb, Widrow, Rosenblatt

    Dose of reality(1966-1974):

  • Simon predicted that in 10 years would have an intelligente machine. Similar predictions in Automatic translation and Chess. Theory of NP-completness. Experiments in Machine evolution(now Genetics alghorytms) (Friedber, 58) was very good founded but could not produced anything. Minsky and Papert Perceptrons (69) (although ironically then same year he discovered the backpropagation (Bryson y Ho))
Sistems based in knowlegment (1969-1979):
  • Dendral , Mycin , HPP, Prospector, Winograd SHDRLU,Shank (there is not sintaxis), frames, Prolog, Planner
  • IA like industry (1980-1988): R1/XCON, project of fiveth generation, shells and machines of Lisp...
Return of Neural Network (1986-now):
  • Hopfield, Rumelhart and Hinton and reduction of the SE
Recents Events (1987-now):
  • Gradual change to technical and far away from rudes(this implicate a degree of maduration and stability) e.g., voice recognition (HMM), uncertainty (Bayes), planning (Tweak), robotics, learning (PAC), etc...

Historical Systems

GPS - Newell, Shaw and Simon
Perceptron - Minsky and Papert Chekers - A. Samuel
MACSYMA AM - D. Lenat
ELIZA - Weisenbaum
Shakey - SRI
SHDRLU - Winogard
MYCIN - E. Shortliffe
Prospector - Duda, Hart
Hearsay II - Erman, Hayes-Roth, Lesser, Reddy
CYC - D. Lenat, R. Guha

Current Systems

PEGASUS - hace reservas de transporte por teléfono [Zue et al. 94]
MARVEL - Expert System in real time that is capable to have a lot of information of Voyager and help to diagnose faults [Schwuttke 92]
Diagnosis System in intern medicine capable of explain with ows reasons to experts [Heckerman 91]
NAVLAB - independence vehicle that can travel in automatic form from part to part of EUA [Pomerlau 93]
Visual System that monitoring the traffic in Paris and recognice crashes in wich it call to emergency services [Koller 94]
DEEP BLUE - paralel machine that plays to chess and the first win to champions of the world[IBM 97]

 

Main - Previous