Metrika članka

  • citati u SCindeksu: [1]
  • citati u CrossRef-u:0
  • citati u Google Scholaru:[=>]
  • posete u poslednjih 30 dana:9
  • preuzimanja u poslednjih 30 dana:1
članak: 1 od 1  
Vojnotehnički glasnik
2015, vol. 63, br. 3, str. 11-28
jezik rada: engleski
vrsta rada: izvorni naučni članak
objavljeno: 01/10/2015
doi: 10.5937/vojtehg63-7529
Feedforward neuronske mreže - Levenberg-Marquardt optimizacija i Optimal Brain Surgeon pruning
General Staff of the Serbian Army, Department for Telecommunication and Informatics (J-6), Centre for Applied Mathematics and Electronics, Belgrade

e-adresa: adanijela@ptt.rs

Sažetak

U radu su opisani obučavanje, testiranje i pruning feedforward neuronske mreže sa jednim skrivenim slojem koji je korišćen za predikciju vokala a. Opisane su Gradient Descent, Gauss-Newton i Levenberg- Marquardt optimizacione tehnike. Optimal Brain Surgeon pruning je primenjen na treniranu mrežu. Kriterijum zaustavljanja je nagla promena normalizovane sume kvadrata grešaka. Struktura feedforward neuronske mreže (FNN) bila je 18 ulaza (četiri za glotalne i 14 za odbirke govora). Rezultati su pokazali da, nakon pruninga, glotalni signal nema uticaja na model za ženskog govornika, dok utiče na predikciju govora kod muškog govornika. U oba slučaja, struktura FNN je redukovana na mali broj parametara.

Ključne reči

Levenberg-Marquardt; analiza govora; pruning; feedforward neuronske mreže

Reference

Azimi-Sadjadi, M.R., Liou, R.-J. (1992) Fast learning process of multilayer neural networks using recursive least squares method. IEEE Transactions on Signal Processing, 40(2): 446-450
Bishop, C.M. (1995) Neural networks for pattern recognition. Oxford: Clarendon Press
Fahlman, S.E. (1988) Fast-learning variation on back propagation: An empirical study. u: The 188 Connectionist Model Schools, San Mateo, Pitsburg, USA, Proceedings of, 38-51
Gavin, H. (2013) The Levenberg-Marquardt method for nonlinear least squares curvefitting problems. Duke University-Department of Civil and Environmental Engineering
Hansen, L.K., Rasmussen, C.E. (1994) Pruning from Adaptive Regularization. Neural Computation, 6(6): 1223-1232
Hassibi, B., Stork, D.G. (1993) Second order derivatives for network pruning: Optimal brain surgeon. u: Hanson S.J., Cowan J.D., Giles C.L. [ur.] Advances in Neural Information Processing Systems, pp. 164-171
Hassibi, B., Stork, D.G., Wolff, G.J. (1993) Optimal brain surgeon and general network pruning. u: IEEE International Conference on Neural Networks, Retrieved from http://systems.caltech.edu/EE/Faculty/babak/pubs/conferences/00298572.pdf doi:10.1109/icnn.1993.298572
Ihler, A. (2013) Linear regression: Gradient descent & stochastic gradient descent
Irvine BREN. ICS University of California, Retrieved from https://www.youtube.com/watch?v=WnqQrPNYz5Q
Kashyap, R. (1980) Inconsistency of the AIC rule for estimating the order of autoregressive models. IEEE Transactions on Automatic Control, 25(5): 996-998
Kwak, Y.T., Hwang, J.W., Yoo, C.J. (2011) A new damping strategy of Levenberg-Marquardt algorithm for Multilayer Perceptrons. Neural Network World, 21(4): 327-340
Larsen, J. (1993) Design of neural networks. Lyngby: Electronic Institute DTH
Levenberg, K. (1944) A method for the solution of certain problems in least squares. Quart Appl Math, 2, pp. 164-168
Ljung, L. (1987) System identification: Theory for the user. Prentice Hall Inc
Marquardt, D.W. (1963) An algorithm for least-squares estimation of nonlinear parameters. SIAM Journal on Applied Mathematics, 11, 431-441
Protić, D.D. (2014) A comparative analysis of Serbian phonemes: Linear and non-linear models. Vojnotehnički glasnik, vol. 62, br. 4, str. 7-37
Ranganathan, A. (2004) The Levenberg-Marquardt algorithm. Retrieved from http://www.ananth.in/docs/lmtut.pdf
Riecke, L., Esposito, F., Bonte, M., Formisano, E. (2009) Hearing Illusory Sounds in Noise: The Timing of Sensory-Perceptual Transformations in Auditory Cortex. Neuron, 64(4): 550-561
Shahin, A.J., Pitt, M.A. (2012) Alpha activity marking word boundaries mediates speech segmentation. European Journal of Neuroscience, 36(12): 3740-3748
Silva, L.M., de Sá, J. M., Alexandre, L.A. (2008) Data classification with multilayer perceptrons using a generalized error function. Neural Networks, 21(9): 1302-1310
Wu, W., Wang, J., Cheng, M., Li, Z. (2011) Convergence analysis of online gradient method for BP neural networks. Neural Networks, 24(1): 91-98
Zinn-Bjorkman, L. (2010) Numerical optimization using Levenberg-Marquardt algorithm. EES-16. LA-UR-11-12010. PowerPoint presentation. PowerPoint presentation