Patent
1994-10-21
1997-01-21
Downs, Robert W.
G06F 1518
Patent
active
055966810
ABSTRACT:
An object of the present invention is to determine the optimal number of neurons in the hidden layers of a feed-forward neural network. The number of the neurons in the hidden layers corresponds to the number of the independent variables of a linear question and the minimum number of the variables required for solving a linear question can be obtained from the rank value in the matrix theory. Therefore, the rank value corresponds to the minimum number of the neurons required for the hidden layers. Accordingly, when the relation between the neural network constructed and trained is memorized in matrix and the rank value is obtained from this matrix, if the number of the neurons in use is larger than the rank value, as it means that redundant neurons exist which correspond to dependent variables, such redundant neurons can be eliminated. In many cases, due to errors in calculation, the diagonal elements of the matrix are not reduced to 0 and consequently the rank value can not be determined. By neglecting the diagonal elements that are smaller than the specific value e, however, the rank value can be estimated within the range of the error.
REFERENCES:
Sartori, M. A., "A Simple Method to Drive Bounds on the Size and to Trim Multilayer Neural Network", IEEE Trans on NN, v. 2, N. 4, 1991, pp. 467-471.
Weigend, A. S., "The effective Dimension of the Space of Hidden Units," IEEE 5th Int. Conf. on NN, Singapore, 1991, pp. 2069-2074.
Caponetto, R. et al, "Genetic Algorithms to select optimal neural Network Topology", 1992 CAS Midwest Symp., pp. 1381-1383.
Ewerbring, L., "Computing the Singular Value Decomposition on the Connection Machine," IEEE Trans. on Computers, v. 39, Iss. 1, 1990, pp. 152-155.
Reed:, "Prunning Algorithms- A Survey", IEEE Transactions on Neural Networks, vol.4, No.5, Sep.1993, pp. 740-747.
Lee, et al: "An NN Based Tone Classifier for Cantonese", Proceedings of 1993 International Joint Conference of Neural Networks, pp. 287-290.
Tamura et al: "Determination of The Number of Redundant Hidden Units in a Three-layered Feed-forward Neural Network", Proceedings of 1993 International Joint Conference on Neural Networks.
Hayashi:, "A Fast Algorithm for the Hidden Units in a Multilayer Perceptron", Proceedings of 1993 International Joint Conference on Neural Networks, pp. 339-342.
Waibel:, "Speech Recognitin Using Time-Delay Neural Networks", Snowbird Conference, 1988, pp. 106-110.
Lippmann: "An Introduction to Computing with Neural Nets", IEEE ASSP Magazine, Apr. 1987, pp. 4-23.
Rumelhart et al: "Parallel Distributed Processing" MIT Press 1986, vol. 1, pp. 322-327.
Downs Robert W.
Katbab A.
Nippondenso Co. Ltd.
LandOfFree
Method of determining an optimal number of neurons contained in does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Method of determining an optimal number of neurons contained in , we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Method of determining an optimal number of neurons contained in will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2331187