240 (4)

240

[46]    5. Jankowski: Neural nctworks — Iheories and upplications. XV KKTOiUF.. Szczyrk, 1992. pp.

626-644

1471 H. Gierasimowicz, S. Jankowski, S. Chachula, i), Pisz: Design of neural <i/d corwertcrs, XIV KKTOiUB, Waplewo 1991. pp. 586-591

(48]    4. Dzicluiski. S. Skonieczny, W. Żydanowicz: Hopfield neural nerwork for opiimal image

filicring. IFAC Symp. IC1C. Sicica 92. Malaga 1992, pp. 489 - 494

[49.1 S. Skoneę-zny, W. Witczak, M. Zajdel: CNN for gray level image Processing. Im. Conf. Comp. Anal. and Paltem Recogu., Budapcst 1993

(50) T. Kacprzak, K. Slot: Somc rcmarks on CNN paramciers determinaliori for fenturc extraction of images. KKTOiUF. Waplewo 1991, pp. 592-597

|5IJ    G. Klimauskas: Neural Warc - User manuał. Neural Ware Inc.. Nalick 1988.

[52]    P. Schmitz: Einfuhrungsbnnd — Backpropagalion. KI - Technik, Regensburg 1989

1531    .-V/. Mougeot. R. Azencott, II. Angeniol: Image compression willi backpropagalion: improvcmcni

of ibe visual rcstoration using differeni cost funetions. Neural Networks. 1991. vol. 4. pp. 467-476

|54J P. Pinceta: Gcncrałization of backpropagalion lo recurrcnt neural nctworks. Phys. Rev. Lcttcrs. 1987. vol. 18. pp. 2229-2232

|55| G. Cybcnko: Approzimation hy superposilion of sigmoidal funciion. Mathcmałics of Control, Signals and Systems, 1989. vol. 2

[561    A. Ivakłtncnko: Polynomial Uieory of complc* Systems. 1EEF. Trans. Syst. Man. Cybern. vol. 1.

pp. 36-1-378

(57| T. Scjnowski. C. Rosenberg: Parallel nctworks Ihai. learn lo prónourtce English tcxt. Complex Systems, I987, vol. I, pp. I45-168

[58] K. Fukushima, N. Wake: Handwrilten alphanumerie cbaracier rccognilion by ihe ncocognilron. IEEE Trans. N. N. vol. 2. I99I. pp. 355-365

[591    A. Rodrif>uez-Vazqnez» R. Dominiguez-Castro. A. Rueda, J. Ilucrlas, li. Sancbcz-Sinencio:

Nonlincar SC neural neiworks for optimization problcms. IEEE Trans. CAS, I990, vol. 37. pp. 384-398

J60| A. Cnlhane. M. Peckerar, C. Marrian: Neural net approach to Ihe discrcle Hartley and Fourier transforms. IEEE Trans. CAS, 1989, vol. 36. pp. 695-703

[611    .V. Gera. J. Sine: A constructivc method for mullivariablc funciion approximauon by multilayer

perceptron. IEEE Trans. N. N.. 1992. voL 3. pp. 621-625

[621    K. Mehrotra, C. Mohan, S. Ranka: Bounds on the number of sninples necdcd for neural

leaming. IEEE Trans. N. N.. 1991, vol. 2, pp. 548-558

[63] A. Michel, ./. Parrei W. Porod: Qualitniive nnulysis of neural nctworks. IEEE Trans. CAS, 1989. vol. 36. pp. 229-243

|64| D. Hnsh, B. Horne: Piogress in supcrviscd neural nctworks. IEEE Signal Processing Magazinc, 1993,January, pp. 8-39

[65]    V. Vapnik, A. Chetronenkis: On the uniform convergence of rclaiive freąucncics of events to their probabililies. Theory of Probability and ics Applications, 1971. vol. 16, pp. 264-280

[66]    5. Son lag: Sigmoids distinguish morę efllciently Ihan heavisidcs. Neural Computations, 1989. vol. 1. pp. 470-472

[67]    Y. Le Cnn. J. Denker, S. Solki: Opiimal Brain Damage. In: „Adyances in Neural Information Processing Systems 2”, ed. D. Touretzky. Morgan Kauffmann. 1990, pp. 598 605