Abstract
In this study we consider a multilayer perceptron network with sigmoidal activation and trained via the backpropagation algorithm. The output of all neurons is collected and a simple linear regression is performed. It is shown that untrained networks with randomly chosen coefficients perform comparably with fully trained networks. This result casts a new light on the role of activation functions, the impact of dimensionality, and the efficacy of training algorithms such as backpropagation.
Similar content being viewed by others
References
Amari, S., Characteristics of randomly connected threshold-element networks and networks systems,Proc. IEEE, vol. 59, 1971.
Carpenter, G., et al., Fuzzy ARTMAP,IEEE Trans. Neural Networks, vol. 3, no. 5, Sept. 1992.
Chiang, C. C. and Fu, H. C., A new simplified second-order multilayer perceptron for improving learning speed and generalization capability, private communication.
Lang, K. K., and Witbrock, M. K., Learning to tell two spirals apart,Proc. 1988 Connectionist Models Summer School, pp. 52–61, Morgan Kaufman, 1988.
Porter, W. A., Using polynomial embeddings for neural network design,IEEE Trans. Neural Networks, vol. 39, no. 6, 1992.
Porter, W. A., Recent progress on higher order moment neural arrays,Circuits Systems Signal Process., vol. 12, no. 2, pp. 309–329, 1993.
Porter, W. A., and Liu, W., Object recognition by a massively parallel 2-D neural architecture,Multidimensional Systems and Signal Processing, vol. 5, pp. 179–201, 1994.
Porter, W. A., and Liu, W., On the performance of higher order moment neural computation,Proc. IEEE SoEastCon, Miami, FL, April 1994.
Porter, W., and Liu, W., Neural network training enchantment, to appear,Circuits, Systems, and Signal Processing, 1996.
Porter, W., and Zhang, X., A nonbinary neural network design,IEEE Trans. Comp., vol. 42, no. 9, pp. 1132–1135, Sept. 1993.
Rozonoer, L., Random logic networks I, II, III, inAutomatic Remote Controls, vols. 5–7, 1969.
Wilensky, G., Analysis of neural network issues,DARPA Neural Network Program Review, Oct. 29–30, 1990.
Author information
Authors and Affiliations
Rights and permissions
About this article
Cite this article
Porter, W.A., Liu, W. & Wu, CY. Auxilliary computations for perceptron networks. Circuits Systems and Signal Process 15, 51–69 (1996). https://doi.org/10.1007/BF01187693
Received:
Revised:
Issue Date:
DOI: https://doi.org/10.1007/BF01187693