Synopsis
Some of the recent work done on layered feed-forward networks is reviewed. First we describe exact solutions for the dynamics of such networks, which are expected to respond to an input by going through a sequence of preassigned states on the various layers. The family of networks considered has a variety of interlayer couplings: linear and nonlinear Hebbian, Hebbian with Gaussian synaptic noise and with various kinds of dilution, and the pseudoinverse (projector) matrix of couplings. In all cases our solutions take the form of layer-to-layer recursions for the mean overlap with a (random) key pattern and for the width of the embedding field distribution. Dynamics is governed by the fixed points of these recursions. For all cases nontrivial domains of attraction of the memory states are found. Next we review studies of unsupervised leaming in such networks and the emergence of orientation-selective cells. Finally the main ideas of three supervised leaming procedures, recendy introduced for layered networks, are oudined. All three procedures are based on a search in the space of intemal representations; one is designed for leaming in networks with fixed architecture and has no associated convergence theorem, whereas the other two are guaranteed to converge but may require expansion of the network by an uncontrolled number of hidden units.
Access this chapter
Tax calculation will be finalised at checkout
Purchases are for personal use only
Preview
Unable to display preview. Download preview PDF.
References
For a recent review of physicists’ contributions see W. Kinzel, Physica Scripta T25, 144 (1989)
E. Domany, R. Meir and W. Kinzel: Europhys. Lett. 2, 175 (1986)
R. Meir and E. Domany: Phys. Rev. Lett. 59, 359 (1987)
R. Meir and E. Domany: Europhys. Lett. 4, 645 (1988)
R. Meir and E. Domany Phys. Rev. A 37, 608 (1988)
R. Meir: J. Phys. (Paris) 49, 201 (1988)
B. Derrida and R. Meir: Phys. Rev. A 38, 3116 (1988)
E. Domany, W. Kinzel and R. Meir: J. Phys. A 22, 2081 (1989)
R. Linsker: Proc. Nat. Acad. Sci. USA 83, 7508–7512 (1986)
R. Linsker: Proc. Nat. Acad. Sci. USA 83, 8390–8394 (1986)
R. Linsker: Proc. Nat. Acad. Sci. USA 83, 8779–8783 (1986)
R. Linsker, in:Computer Simulation in Brain Science, edited by R. Coltrili (Cambridge University Press, Cambridge 1988)
R. Linsker: IEEE Computer (March 1988) 105–117
9.12R. Linsker: to be published in the proceedings of the 1988 Denver Conference on Neural Information Processing Systems (Morgan Kauffman)
R. Meir and E. Domany: Phys. Rev. A 37, 2660 (1988)
T. Grossman, R. Meir, and E. Domany: Complex Systems, 2, 555 (1988)
P. Rujan and M. Marchand: Complex Systems 3, 229 (1989)
M. Mézard and J.P. Nadal: J. Phys. A 22, 2191 (1989)
See for example J.E. Hopcroft and R.L. Mattson, Synthesis of Minimal Threshold Logic Networks, IEEE Trans. Electronic Computers, EC-14, 552 (1965)
J.J. Hopfield: Proc. Natl. Acad. USA 79, 2554 (1982)
DJ. Amit, H. Gutfreund, and H. Sompolinsky: Ann. Phys. 173, 30 (1987)
H. Sompolinsky, in: Heidelberg Colloquium on Glassy Dynamics edited by J.L. van Hemmen and I. Morgenstem, Lecture Notes in Physics Vol. 275 (Springer, Berlin, Heidelberg 1987)
B. Derrida, E. Gardner, and A. Zippelius: Europhys. Lett. 4, 167 (1987)
See for example E. Domany, J. Stat. Phys. 51, 743 (1988)
9.23See H. Ritter, K. Obermayer, K. Schulten, and J. Rubner, this volume, Chap. 8
E. Gardner: J. Phys. A 21, 257 (1988)
E. Gardner and B. Derrida, J. Phys. A 21, 271 (1988)
W. Krauth and M. Opper: J. Phys. A 22, L519 (1989)
L Kanter and H. Sompolinsky: Phys. Rev. A 35, 380 (1987)
A.C.C. Coolen, J.J. Denier van der Gon, and Th.W. Ruijgrok: Proc. nEuro88
A.C.C. Coolen, H J.J. Jonker, and Th.W. Ruijgrok, Utrecht preprint (1989)
M. Opper, J. Kleinz, H. Köhler, and W. Kinzel: J. Phys. A 22, L407 (1989)
T.B. Kepler and L.F. Abbott: J. Phys. (Paris) 49, 1657 (1988)
H. Homer, D. Bormann, M. Frick, H. Kinzelbach, and A. Schmidt: Z. Phys. B 76,381 (1989)
I. Kanter: Phys. Rev. A 40, 2611 (1989)
S. Diedrich and M. Opper: Phys. Rev. Lett. 58, 949 (1987)
E. Gardner, N. Stroud, and DJ. Wallace: J. Phys. A 22, 2019 (1989)
W. Krauth and M. Mézard: J. Phys. A 21, L745 (1987)
P. Peretto: Neural Networks 1, 309–322 (1988)
J.F. Fontanari and R. Meir: Caltech preprints (1989)
L.F. Abbott and T.B. Kepler: J. Phys. A 22, L711 (1989)
FJ. Pineda: Phys. Rev. Lett. 59, 2229 (1987)
M. Opper: Europhys. Lett. 8, 389 (1989)
J.A. Hertz, G.L Thorbergson, and A. Krogh: Physica Scripta T25, 149 (1989)
W. Kinzel and M. Opper: this volume. Chap.4
For reviews see: J.D. Cowan and D.H. Sharp, Quarterly Reviews of Biophysics, 21 365 (1988)
R.P. Lippmann, IEEE ASSP Magazine, 4, 4 (1987)
T. Kohonen: Self Organization and Associative Memory (Springer, Beriin, Heidelberg 1984)
D.E. Rumelhart and J.L. McClelland: Parallel Distributed Processing: Explorations in the Microstructure of Cognition, 2 vols. (MTT Press, Cambridge, Mass. 1986)
M. Minsky and S. Papert:Perceptrons, expanded edition (MIT Press, Cambridge, Mass. 1988)
D.E. Rumelhart, G.E. Hinton, and R J. Williams: in Parallel Distributed Processing: Explorations in the Microstructure of Cognition edited by D.E. Rumelhart and J.L. McClelland, (MIT Press, Cambridge, Mass. 1986) Vol. 1, p. 318
Y. Le Cun: Proc. Cognitiva, 85, 593 (1985)
PJ. Werbos, Ph.D. thesis. Harvard University (1974)
D.B. Parker, MIT Technical Report TR-47 (1985)
M. Abeles: Local Cortical Circuits (Springer, Berlin, Heidelberg 1982)
H. Sompolinsky: Phys. Rev. A 34, 2571 (1986)
J.L. van Hemmen and R. Kühn, Phys. Rev. Lett. 57, 913 (1986)
J.L. van Hemmen, Phys. Rev. A 36, 1959 (1987)
W. Krauth, M. Mézard, and J.P. Nadal: Complex Systems, 2, 387 (1988)
W.A. Litüe: Math. Biosci. 19, 101 (1975)
S. Amari and K. Maginu: Neural Networks, 1, 63 (1988)
See for example W. Feller, An Introduction to Probability Theory and its Applications (Wiley, New York 1966) Vol. 4 p. 256
9.54It is very important to realize that the embedding fields Hi and Hj are not independent. Their correlation is M N. This correlation gives rise to the layer-to-layer recursive variation of the width parameter A which in tum, causes the appearance of non trivial domains of attraction
B. Derrida: J. Phys. A 20, L72I (1987)
W. Kinzel: Z. Physik B 60, 205 (1985)
W. Krauth, J.P. Nadal, and M. Mézard: J. Phys. A: Math. Gen. 21, 2995 (1988)
J. Kleinz: diploma thesis, Justus-Liebig University Glessen (1988)
For a recent review see D. H. Hubel, Los Alamos Science 16, 14 (1988)
D. Kämmen and A. Yuille: Biol. Cybem. 59, 23 (1988)
P. Huben Ann. Statistics 13, 435 (1985)
E. Oja: J. Math. Biol. 15, 267 (1982)
R. E. Blahut: Principles and Applications of Information Theory, (Addison-Wesley, 1987)
B. Widrow and R. Winter: Computer 21, 25 (1988)
F. Rosenblatt: Psych. Rev. 62, 386 (1958)
F. RosenblattPrinciples of Neurodynamics (Spartan, New York 1962)
P.M. Lewis and C.L. Coates: Threshold Logic (Wiley, New York 1967)
J. Denker, D. Schwartz, B. Wittner, S. Solla, J.J. Hopfield, R. Howard, and L. Jackel: Complex Systems 1, 877–922 (1987)
T. Grossman, Complex Systems 3, 407 (1989)
T. Grossman, in Advances in Neural Information Processing Systems 2, edited by D. Touretzky (Morgan Kaufman, San Mateo 1990) p. 516
R. Rohwer, in Advances in Neural Information Processing Systems 2, edited by D. Touretzky (Morgan Kaufman, San Mateo 1990) p. 538)
A. Krogh, G.I. Thorbergsson, and J.A. Hertz, ibid, p. 773
D. Saad and E. Marom, Complex Systems 4, 107 (1990)
D. Nabutovsky, T. Grossman, and E. Domany: Complex Systems 4, 519 (1990)
Editor information
Editors and Affiliations
Rights and permissions
Copyright information
© 1995 Springer-Verlag Berlin Heidelberg
About this chapter
Cite this chapter
Domany, E., Meir, R. (1995). Layered Neural Networks. In: Domany, E., van Hemmen, J.L., Schulten, K. (eds) Models of Neural Networks I. Physics of Neural Networks. Springer, Berlin, Heidelberg. https://doi.org/10.1007/978-3-642-79814-6_9
Download citation
DOI: https://doi.org/10.1007/978-3-642-79814-6_9
Publisher Name: Springer, Berlin, Heidelberg
Print ISBN: 978-3-642-79816-0
Online ISBN: 978-3-642-79814-6
eBook Packages: Springer Book Archive