论文信息 - A Simple Strategy to Prune Neural Networks with an Application to Economic Time Series

A Simple Strategy to Prune Neural Networks with an Application to Economic Time Series

A major problem in applying neural networks is specifying the sizeof the network. Even for moderately sized networks the number ofparameters may become large compared to the number of data. In thispaper network performance is examined while reducing the size of thenetwork through the use of multiple correlation coefficients andgraphical analysis of network output per hidden layer cell andinput layer cell.

Johan F. Kaashoek | Herman K. van Dijk | H. V. Dijk | J. Kaashoek

[1] Anders Krogh,et al. Introduction to the theory of neural computation , 1994, The advanced book program.

[2] F. Takens. Detecting strange attractors in turbulence , 1981 .

[3] Herman K. van Dijk,et al. On Bayesian routes to unit roots , 1991 .

[4] G. Schwarz. Estimating the Dimension of a Model , 1978 .

[5] H. White,et al. There exists a neural network that does not make avoidable mistakes , 1988, IEEE 1988 International Conference on Neural Networks.

[6] H. Theil. Principles of econometrics , 1971 .

[7] Michael C. Mozer,et al. Skeletonization: A Technique for Trimming the Fat from a Network via Relevance Assessment , 1988, NIPS.

[8] H. White. Some Asymptotic Results for Learning in Single Hidden-Layer Feedforward Network Models , 1989 .

[9] R. Devaney. An Introduction to Chaotic Dynamical Systems , 1990 .

[10] Heekuck Oh,et al. Neural Networks for Pattern Recognition , 1993, Adv. Comput..