This Article 
 Bibliographic References 
 Add to: 
A Performance Model for Multilayer Neural Networks in Linear Arrays
December 1994 (vol. 5 no. 12)
pp. 1322-1328

An analytical model is presented for assessing the performance of multilayer neuralnetworks implemented in linear arrays. Metrics to assess latency, throughput rate, andcomputational and input-output bandwidth are developed. These metrics demonstrate arich and complex interaction between the performance of the hardware and the numberand relative dimensions of the layers in a network. Practical illustration of the use ofthese metrics is demonstrated for a two-hidden-layer network.

[1] S. Y. Kung and J. N. Hwang, "A unifying algorithm/architecture for artificial neural networks,"Int. Conf. Application Specific Signal Processors, 1990, pp. 2505-2508.
[2] D. Hammerstrom, "A highly parallel digital architecture for neural network emulation," inVLSI for Artificial Intelligence and Neural Nerworks, J. G. Delgado-Frias and W. Moore, Eds. New York: Plenum, 1991, pp. 357-366.
[3] D. J. Myers, J. M. Vincent, and D. A. Orrey, "A VLSI architecture for implementing neural networks with on-chip backpropagation learning," inNeural Networks for Vision, Speech and Natural Language, R. Linggard, D. J. Myers and C. Nightingale, Eds. London, UK: Chapman&Hall, 1992, pp. 312-329.
[4] D. E. Rumelhart, G. E. Hinton, and R. J. Williams, "Learning internal representation by error propagation,"Parallel Distributed Processing: Explorations in the Microstructure of Cognition, vols. 1 and 2. Cambridge, MA: MIT Press, 1986.

Index Terms:
Index Termsfeedforward neural nets; systolic arrays; performance evaluation; multilayer perceptrons;performance model; multilayer neural networks; linear arrays; latency; throughput rate;input-output bandwidth; performance; two-hidden-layer network
D. Naylor, S. Jones, "A Performance Model for Multilayer Neural Networks in Linear Arrays," IEEE Transactions on Parallel and Distributed Systems, vol. 5, no. 12, pp. 1322-1328, Dec. 1994, doi:10.1109/71.334906
Usage of this product signifies your acceptance of the Terms of Use.