Autosoft Journal

Online Manuscript Access

Abstracting Uncertain Knowledge: Case for Neural Nets Application



Neural networks have been used successfully in practical applications where human expertise exists but no clear rules are known. There is a more difficult case when the acquisition of labelled data points is very expensive, such as in labelling of ground data to match satellite images in geographic information systems.In fact, the dependence of neural networks on large volumes of training data result in the neural solution, producing more inconsistent results over a number of trials using the same data, but different initialisations of the weights.We present our method of generating IF-THEN rules expressing the trained neural networkaposs behaviour. By using the causal index on characteristic input patterns, we produce a list of inputs which were significant in reaching the decision made and a well-ordered sequence of rules governing this decision. This method correctly produced rules for 94percnt of the decisions made by a sample network.The principle of selecting the next most likely decision (that the network could have made) brings forth the question of specificity of the ensuing procedure. The extent to which each rule ldquooutweighsrdquo its successor implies the degree of our confidence in the correctness of the final decision. Viewing causal indices as degrees of possibility of rules relevance permits using the machinery of formal specificity measures to capture this notion quantitatively. We outline the methodology, which has a well-defined axiomatic basis, and leads to a parametrized family of specificity functions.



Total Pages: 13
Pages: 365-377


Manuscript ViewPdf Subscription required to access this document

Obtain access this manuscript in one of the following ways

Already subscribed?

Need information on obtaining a subscription? Personal and institutional subscriptions are available.

Already an author? Have access via email address?


Volume: 1
Issue: 4
Year: 1995

Cite this document


Rumelhart D.E. Parallel Distributed Processing

Gedeon T.D. Proc. Int. Joint Conf. on Neural Networks

Gedeon T.D. Proceedings Brazil-Japan Joint Symposium on Fuzzy Systems

Gallant S.I. Communications of the ACM 31.2 (1988)

Towell G.G. Machine Learning

Craven M.W. 10th International Machine Learning Workshop

Fu L.M. International Joint Conference on Neural Networks

Yoda M. International Joint Conference on Neural Networks

Hora N. Proc. IEEE

Dubois D. Possibility Theory

Ramer A. Fuzzy Sets and Systems

Yager, R. “On measuring specificity.” Tech. Rep., Iona College, New Rochelle, NY, 1990.

Hardy G. Inequalities

Ramer, A., R. Yager. “Specificity as uncertainty metric.” 3rd IPMU - Int. Conf. Information Proc. and Management of Uncertainty, Madrid, 1992.

Dubois D. Fuzzy Sets and Syst. 24.2 (1987)

Ramer A. Int. J. Gen. Syst.

Slade P. New Trends in Neural Computation.


ISSN PRINT: 1079-8587
ISSN ONLINE: 2326-005X
DOI PREFIX: 10.31209
10.1080/10798587 with T&F
IMPACT FACTOR: 0.652 (2017/2018)
Journal: 1995-Present


TSI Press
18015 Bullis Hill
San Antonio, TX 78258 USA
PH: 210 479 1022
FAX: 210 479 1048