1990-04-16
1991-12-10
Fleming, Michael R.
395 77, G06F 1518, G06F 944
Patent
active
050724061
DESCRIPTION:
BRIEF SUMMARY
TECHNICAL FIELD
The invention relates generally to artificial intelligence, which is a subfield of computer science, and more specifically to a computer system and process of operating a computer that has a memory for storing knowledge, analyzes input signals, improves its efficiency by changing and extending the knowledge in its memory, and produces output signals on the basis of the changed and extended knowledge. The input signals can be generated by keyboards or sensors and the output signals can control printers, displays, or the movements of robots.
BACKGROUND ART
Conventional computer systems including robots are not able to improve their efficiency automatically by substantially changing and extending the programs or the knowledge in their memories (Feigenbaum, E. A., und McCorduck, P., The Fifth Generation: Artificial Intelligence and Japan's Computer Challenge to the World, Addison-Wesley, Menlo Park, Calif., 1983, p. 237). Such changes and extensions of the programs or the knowledge in machines are conventionally executed by human experts on the basis of an analysis of a field of application, that is, the programs or the knowledge are manually developed on the basis of a manual analysis of the field of application.
The ordinary use of conventional computer systems is that programs, which are compositions of elementary complete instructions, and data, which are to be processed by the programs, are loaded into their memories by means of input units, that the programs are applied to the data by means of processors, and that the final results are transmitted to output units.
The field of expert systems relates to development of computer systems having special knowledge in a field of application (Hayes-Roth, F., Waterman, D. A., and Lenat, D. B., Building Expert Systems, Addison-Wesley, London, 1983). An expert system consists of a knowledge base and an inference machine. The knowledge base contains knowledge in a field of application and the inference machine methods for applying this knowledge. The knowledge consists of facts in the field of application and heuristic rules containing experiental knowledge. The knowledge base and the inference machine are manually developed on the basis of a manual analysis of the field of application. In tightly restricted fields of application the efficiency of expert systems is comparable to that of human experts (ibid., p. 38). Two well-known expert systems are MYCIN and R1. MYCIN diagnoses blood deseases and offers advice for their therapy (Shortliffle, E. H., MYCIN: Computer-Based Medical Consultations, American Elsevier, New York, 1976). R1 generates configurations for the VAX computer system of Digital Equipment Corporation on the basis of customer orders. These configurations consist of diagrams which contain the spacial relations of the components in the orders (McDermott, J., "R1: A Rule-Based Configurer of Computer Systems", Technical Report CMU-CS-80-119, Computer Science Department, Carnegie-Mellon University, Pittsburg, Pa. 1980).
The field of machine learning has the object to develop computer systems that change and extend the knowledge in their memories. Some well-known learning systems are briefly described subsequently. Systems using a method called explanation-based learning transform a manually given inefficient definition of a concept to be learned into an efficient definition on the basis of a manually given description of an example, a manually given theory of a field of application, and manually given criteria specifying efficient expressions in which the concept to be learned shall be represented (Mitchell, T. M., Keller, R. M., and Cedar-Cabelli, S. T., "Explanation-based generalization: A unifying view", Machine Learning, Vol. 1, No. 1, 1986, pp. 47-80). The LP system learns new methods for solving equations by a procedure called precondition analysis (Silver, B., "Precondition analysis: Learning control information", in R. S. Michalski, J. G. Carbonell, and T. M. Mitchell (Hrsg.), Machine Learning: An Artificial Intelligence Approac
REFERENCES:
patent: 4591983 (1986-03-01), Bennett et al.
patent: 4670848 (1987-06-01), Schramm
patent: 4853873 (1989-08-01), Tsuji et al.
patent: 4868763 (1989-09-01), Masui et al.
Davis George
Fleming Michael R.
LandOfFree
Self-developing computer system does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Self-developing computer system, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Self-developing computer system will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-1045856