QuoteRef: lenaDB10_1982

topics > all references > ThesaHelp: references i-l

references i-l
heuristic-based systems
definition by example
attribute-value pairs as information
evolutionary systems
man-machine symbiosis


Lenat, D.B., "The nature of heuristics", Artificial Intelligence, 19, 2, October 1982, pp. 189-249. Google

190 ;;Quote: a heuristic domain must be observable with sufficient continuity and stability to make heuristics useful
191 ;;Quote: very few fields admit automatic data acquisition for developing heuristics
191 ;;Quote: developing heuristics by analogy appears to be better than specialization and generalization; e.g., find examples before proofs
199 ;;Quote: a heuristic is a contingent piece of guiding knowledge; limited scope
199+;;Quote: a maxim like 'compose two operations and study the result' generates too many possibilities; use an heuristic instead
200 ;;Quote: in almost all cases, the discoveries made by AM were unexpected or unknown by the author; not obvious from AM's heuristics
200 ;;Quote: AM selected a slot of some concept and worked to fill it; started with 100 concepts, about 20 slots for each (e.g., Examples, Conjectures)
201 ;;Quote: to accomplish a task, AM obeyed relevant heuristics; they caused entries to be filled in and defined new concepts and tasks
204 ;;Quote: while most heuristic search programs use heuristics to prune, AM used heuristics to create the search space
204+;;Quote: almost half of the concepts proposed by AM's heuristics were interesting
204 ;;Quote: each AM heuristic applied to a concept and all its specializations; so interpreter only considered potentially relevant heuristics
207 ;;Quote: of AM's first 200 concepts, 124 were judged acceptable; of the next 300 concepts, only 29 were acceptable; AM started with 114 concepts
220 ;;Quote: the heuristics used by AM clustered at the root and leaves of the concept specialization tree
225 ;;Quote: EURISKO rests on not differentiating between meta-level heuristics about heuristics from object-level heuristics
231 ;;Quote: the slot of an AM unit is like an attribute/value pair where the value need not be stored explicitly
236 ;;Quote: EURISKO uses a large number of specialized slots; allows a condition to be an atom in some slot instead of bulky, if/then rules
241 ;;Quote: EURISKO won the 1981 Traveler game; first attempt at a miniatures battle game (several hundred rules and constraints)
242 ;;Quote: for the 1981 Traveler game, Lenat culled the heuristics generated by EURISKO every 12 hours or so; neither could have won alone

Related Topics up

ThesaHelp: references i-l (342 items)
Topic: heuristic-based systems (35 items)
Topic: definition by example (26 items)
Topic: attribute-value pairs as information (53 items)
Topic: evolutionary systems (38 items)
Topic: man-machine symbiosis (46 items)

Collected barberCB 4/88
Copyright © 2002-2008 by C. Bradford Barber. All rights reserved.
Thesa is a trademark of C. Bradford Barber.