DATA-DEPENDENT WEIGHT INITIALIZATION IN THE HOPFIELD NEURAL NETWORK CLASSIFIER: APPLICATION TO NATURAL COLOUR IMAGES

R. Sammouda

References

  1. [1] C.M. Bishop, neural networks for pattern recognition, Chapter7 (Oxford University Press, 1995), 253–294.
  2. [2] F.V.D. Berg, particle swarm weight initialization in multi-layerperceptron artificial neural networks, IEEE Swarm IntelligenceSymposium Proceeding, 2003, 182–186.
  3. [3] S. Devidhruba, C. Pandashyam, & S. Pattnaik, initializingartificial neural networks by genetic algorithm to calculate theresonant frequency of single shorting post rectangular patch an-tenna, IEEE Antennas and Propagation Society InternationalSymposium, 3, 2003, 144–147.
  4. [4] L.N. de Castro, E.M. Iyoda, F.J.V. Zuben, & R. Gudwin,Feedforward neural network initialization: An evolutionaryapproach, 7th Brazilian Symposium on Neural Networks Pro-ceeding, 1998, 43–48.
  5. [5] R. Sammouda, N. Niki, & H. Nishitani, A comparison ofhopfield neural network and boltzmann machine in segmentingMR images of the brain, IEEE Transactions on Nuclear Science,43 (6), 1996, 349–356.
  6. [6] M. Sammouda, R. Sammouda, N. Niki, et al., Liver cancerdetection system based on the analysis of digitized colour imagesof tissue samples obtained using needle biopsy, InternationalJournal of Information Visualization, 1, 2002, 130–138.
  7. [7] M. Fernandez-Redondo & C. Hernandez-Espinose, Weight ini-tialization methods for multilayer feedforward, European Sym-posium on Artificial Neural Networks Proceeding, 2001, 119–124.
  8. [8] Y.K. Kim & J.B. Ra, Weight value initialization for improvingtraining speed in the backpropagation network, Proceeding ofInternational Joint Conference on Neural Networks, 3, 1991,2396–2401.
  9. [9] G. Li, H. Alnuweiri, & Y. Wu, Acceleration of backpropagationsthrough initial weight pre-training with delta rule, IEEEInternational Conference on Neural Networks Proceeding, 1,1993, 580–585.
  10. [10] H. Yoon, C. Bae, & B. Min, Neural networks using modifiedinitial connection strengths by the importance of feature ele-ments, International Joint Conference on Systems, Man andCybernetics Proceeding, 1, 1995, 458–461
  11. [11] G.P. Drago & S. Ridella, Statistically controlled activationweight initialization (SCAWI), IEEE Transactions on NeuralNetworks, 3 (4), 1992, 627–631.

Important Links:

Go Back