Abstract
The inherent fault tolerance of artificial neural networks (ANNs) is usually assumed, but several authors have claimed that ANNs are not always fault tolerant and have demonstrated the need to evaluate their robustness by quantitative measures. For this purpose, various alternatives have been proposed. In this paper we show the direct relation between the mean square error (MSE) and the statistical sensitivity to weight deviations, defining a measure of tolerance based on statistical sentitivity that we have called Mean Square Sensitivity (MSS); this allows us to predict accurately the degradation of the MSE when the weight values change and so constitutes a useful parameter for choosing between different configurations of MLPs. The experimental results obtained for different MLPs are shown and demonstrate the validity of our model.
This is a preview of subscription content, log in via an institution to check access.
Similar content being viewed by others
Explore related subjects
Discover the latest articles, news and stories from top researchers in related subjects.References
Segee, B. E. and Carter, M. J.: Comparative fault tolerance of parallel distributed processing networks, IEEE Trans on Computers 43(11) (1994), 1323–1329.
Pathak, D. S. and Koren, I.: Complete and partial fault tolerance of feedforward neural nets, IEEE Trans. on Neural Networks 6(2) (1995), 446–456.
Stevenson, M., Winter, R. and Widrow, B.: Sensitivity of neural networks to weight errors, IEEE Trans. on Neural Networks 1(1) (1990), 71–80.
Alippi, C., Piuri, V. and Sami, M.: Sensitivity to errors in artificial neural networks: a behavioral approach, In: Proc. IEEE Int. Symp. on Circuits & Systems, pp. 459–462, May 1994.
Edwards, P. J. and Murray, A. F.: Fault tolerance via weight-noise in analogue VLSI implementations – a case study with EPSILON, IEEE Proc. on Circuits and Systems II: Analog and Digital Signal Processing 45(9) (1998), 1255–1262.
Edwards, P. J. and Murray, A. F.: Can deterministic penalty terms model the effects of synaptic weight noise on network fault-tolerance?, Int. Journal of Neural Systems 6(4) (1995), 401–416.
Choi, J. Y. and Choi, C.: Sensitivity analysis of multilayer perceptron with differentiable activation functions, IEEE Trans. on Neural Networks 3(1) (1992), 101–107.
Sudkamp, T. and Hammell, R.: Interpolation, completion and learning fuzzy rules, IEEE Trans. on Systems, Man & Cybernetics 24(2) (1994), 332–342.
Wang, L.: Adaptive Fuzzy Systems and Control. Design and Stability Analysis, Englewood Cliffs, Prentice Hall, 1994.
Bernier, J. L., Ortega, J. and Prieto, A.: A modified backpropagation algorithm to tolerate weight errors, Lecture Notes in Computer Science 1240, (1997), Springer-Verlag, 763–771.
Author information
Authors and Affiliations
Rights and permissions
About this article
Cite this article
Bernier, J.L., Ortega, J., Rodrìguez, M.M. et al. An Accurate Measure for Multilayer Perceptron Tolerance to Weight Deviations. Neural Processing Letters 10, 121–130 (1999). https://doi.org/10.1023/A:1018733418248
Issue Date:
DOI: https://doi.org/10.1023/A:1018733418248