A neural network expert system with confidence measurements

  • Stephen I. Gallant
  • Yoichi Hayashi
10. Neural Networks
Part of the Lecture Notes in Computer Science book series (LNCS, volume 521)


We show two ways to attach confidence measurements to MACIE, a connectionist expert system model. One technique involves multiple generation of neural network knowledge bases and parallel execution of expert system inference engines. Required modifications for forward and backward chaining and explanations by If-Then rules with confidence measurements are given.

In the second approach we show that inferences made by the MACIE connectionist expert system with partial information have the same confidence as inferences made with full information under certain standard probability models, including distribution-free Valiant-style learning.


confidence MACIE neural network expert system Valiant model 


Unable to display preview. Download preview PDF.

Unable to display preview. Download preview PDF.


  1. [1]
    Frydenberg, M. & Gallant, S. I. Fuzziness and Expert System Generation. International Conference on Information Processing and Management of Uncertainty in Knowledge-Based Systems, Paris, France, June 30–July 4, 1986. Extended paper reprinted in Uncertainty in Knowledge-Based Systems, B. Bouchon & R. Yager Editors, Springer-Verlag: Berlin 1987.Google Scholar
  2. [2]
    Gallant, S. I. Connectionist Expert Systems. Communications of the ACM, Vol. 31, Number 2, Feb. 1988, 152–169. (Japanese translation in Neurocomputer, published by Nikkei Artificial Intelligence, pg. 114–136, 1988.)CrossRefGoogle Scholar
  3. [3]
    Gallant, S. I. Example-Based Knowledge Engineering with Connectionist Expert Systems. Proc. IEEE MIDCON, Dallas, Texas, August 30-September 1, 1988, 32–37.Google Scholar
  4. [4]
    Gallant, S. I. A Connectionist Learning Algorithm With Provable Generalization and Scaling Bounds. Neural Networks, Vol.3, pp. 191–201, 1990.CrossRefGoogle Scholar
  5. [5]
    Gallant, S. I. Automated Generation of Expert Systems for Problems Involving Noise and Redundancy. AAAI Workshop on Uncertainty in Artificial Intelligence, Seattle, Washington, July 10–12, 1987. Pg. 212–221.Google Scholar
  6. [6]
    Gallant, S. I. Bayesian Assessment of a Connectionist Model For Fault Detection. AAAI Fourth Workshop on Uncertainty in Artificial Intelligence, St. Paul, MN, August 19–21, 1988. Pg. 127–135.Google Scholar
  7. [7]
    Gallant, S. I. Perceptron-Based Learning Algorithms. IEEE Transactions on Neural Networks, Vol. 1, No. 2, June 1990, 179–192.CrossRefGoogle Scholar
  8. [8]
    Hayashi, Y. A Neural Expert System with Automated Extraction of Fuzzy If-Then Rules and Its Application to Medical Diagnosis. In Touretzky, D.S. and Lippman, R. (eds.) Advances in Neural Information Processing Systems, Vol. 3, San Mateo, CA: Morgan Kaufmann, 1991 (in press).Google Scholar
  9. [9]
    Hayashi, Y. & Nakai, M. Reasoning Methods Using a Fuzzy Production Rule with Linguistic Relative Importance in an Antecedent. (in Japanese) Trans. IEE of Japan, Vol. 109-C, No. 9, Sept. 1989, 661–668.Google Scholar
  10. [10]
    Hayashi, Y. & Nakai, M. Automated Extraction of Fuzzy IF-THEN Rules Using Neural Networks. (in Japanese) Trans. IEE of Japan, Vol. 110-C, No. 3, March 1990, 198–206.Google Scholar
  11. [11]
    Hayashi, Y. & Imura, A. Fuzzy Neural Expert System with Automated Extraction of Fuzzy If-Then Rules from a Trained Neural Network. Proceedings of the First Int. Symposium on Uncertainty Modeling and Analysis, pp. 489–494, Maryland, Dec. 3–5, 1990.Google Scholar
  12. [12]
    Hisdal, E. Are Grades of Membership Probabilities? First World Congress on Fuzzy Sets and Systems (IFSA '85), Palma de Mallorca, July 1–6, 1985.Google Scholar
  13. [13]
    Hisdal, E. Infinite-Valued Logic Based on Two-Valued Logic and Probability. International Journal of Man-Machine Studies. Part 1.1: Vol. 25 (1986) 89–111, Part 1.2: Vol. 25 (1986) 113–138, Part 1.3–Part 1.6: to appear.Google Scholar

Copyright information

© Springer-Verlag Berlin Heidelberg 1991

Authors and Affiliations

  • Stephen I. Gallant
    • 1
  • Yoichi Hayashi
    • 2
  1. 1.ICANNCambridgeUSA
  2. 2.Department of Computer & Information ScienceIbaraki UniversityIbarakiJapan

Personalised recommendations