Data processing: artificial intelligence – Machine learning
Reexamination Certificate
2000-04-12
2003-02-18
Black, Thomas (Department: 2121)
Data processing: artificial intelligence
Machine learning
C706S013000, C706S014000
Reexamination Certificate
active
06523016
ABSTRACT:
BACKGROUND OF THE INVENTION
Recent years have witnessed a significant progress in the development and applications of machine learning methods, in particular, in scaling them up to cope with large datasets (e.g., Clark and Niblett, 1989; Cohen, 1995; Dietterich, 1997; T. Mitchell, 1997; Michalski, Bratko and Kubat, 1998). There has also been a significant progress in the area of evolutionary computation (e.g., Koza, 1994; Michalewicz, 1996; Baeck, Fogel and Michalewicz, 1997; Banzhaf et al., 1998).
All conventional methods of evolutionary computation draw inspiration from the principles of Darwinian evolution in which basic operators are mutation, crossover (recombination), and selection of the fittest. These operators are very simple and domain-independent; thus, they can be employed without knowing a model of the problem domain (e.g., Holland, 1975; Goldberg, 1989; Michalewicz, 1996; M. Mitchell, 1996). Consequently, the Darwinian-type evolutionary computation has been applied to a wide range of problems, such as various kinds of optimization and search problems, automatic programming, engineering design, game playing, machine learning, pattern recognition, and evolvable hardware.
The Darwinian-type evolution is, however, semi-blind: the mutation is a random modification of the current solution; the crossover is a semi-random recombination of two solutions; and the selection (survival) of the fittest is a form of parallel hill climbing. In this type of evolution, individuals do not pass lessons learned from their experience to the next generation. Consequently, computational processes based on Darwinian evolution are not very efficient. Low efficiency has been the major obstacle in the application of evolutionary computation to very complex problems.
REFERENCES:
patent: 5136686 (1992-08-01), Koza
patent: 5140530 (1992-08-01), Guha et al.
patent: 5148513 (1992-09-01), Koza et al.
patent: 5222192 (1993-06-01), Shaefer
patent: 5245696 (1993-09-01), Stork et al.
patent: 5255345 (1993-10-01), Shaefer
patent: 5343554 (1994-08-01), Koza et al.
patent: 5390282 (1995-02-01), Koza et al.
patent: 5435309 (1995-07-01), Thomas et al.
patent: 5651099 (1997-07-01), Konsella
patent: 5742738 (1998-04-01), Koza et al.
patent: 5857462 (1999-01-01), Thomas et al.
patent: 5867397 (1999-02-01), Koza et al.
patent: 6058385 (2000-05-01), Koza et al.
patent: 6169981 (2001-01-01), Werbos
patent: 6360191 (2002-03-01), Koza et al.
Al science vs. applications: accomplishments, failures and long-term goals, Michalski, R.S.; Computers and Communications, 1991. Conference Proceedings., Tenth Annual International Phoenix Conference on , 1991, pp. 846-847.*
Data-driven constructive induction in AQ17-PRE: A method and experiments, Bloedorn, E.; Michalski, R.S.; Tools for Artificial Intelligence, 1991. TAI '91., Third International Conference on , 1991, pp. 30-37.*
Learning textural concepts through multilevel symbolic transformations, Bala, J.W.; Michalski, R.S.; Tools for Artificial Intelligence, 1991. TAI '91., Third International Conference on , 1991, pp. 100-107.*
A method for partial-memory incremental learning and its application to computer intrusion detection, Maloof, M.A.; Michalski, R.S.; Tools with Artificial Intelligence, 1995. Proceedings., Seventh International Conference on , 1995, pp. 392-397.*
Experimental validations of the learnable evolution model, Cervone, G.; Kaufman, K.K.; Michalski, R.S.; Evolutionary Computation, 2000. Proceedings of the 2000 Congress on, vol.: 2, 2000 pp.: 1064-1071 vol. 2.
Black Thomas
George Mason University
Grossman David G.
Holmes Michael B.
LandOfFree
Learnable non-darwinian evolution does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Learnable non-darwinian evolution, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Learnable non-darwinian evolution will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3158670