The Road to Chaos by Time-Asymmetric Hebbian Learning in Recurrent Neural Networks

This letter aims at studying the impact of iterative Hebbian learning algorithms on the recurrent neural network's underlying dynamics. First, an iterative supervised learning algorithm is discussed. An essential improvement of this algorithm consists of indexing the attractor information items by means of external stimuli rather than by using only initial conditions, as Hopfield originally proposed. Modifying the stimuli mainly results in a change of the entire internal dynamics, leading to an enlargement of the set of attractors and potential memory bags. The impact of the learning on the network's dynamics is the following: the more information to be stored as limit cycle attractors of the neural network, the more chaos prevails as the background dynamical regime of the network. In fact, the background chaos spreads widely and adopts a very unstructured shape similar to white noise. Next, we introduce a new form of supervised learning that is more plausible from a biological point of view: the network has to learn to react to an external stimulus by cycling through a sequence that is no longer specified a priori. Based on its spontaneous dynamics, the network decides on its own the dynamical patterns to be associated with the stimuli. Compared with classical supervised learning, huge enhancements in storing capacity and computational cost have been observed. Moreover, this new form of supervised learning, by being more respectful of the network intrinsic dynamics, maintains much more structure in the obtained chaos. It is still possible to observe the traces of the learned attractors in the chaotic regime. This complex but still very informative regime is referred to as frustrated chaos.

[1]  J. Piaget.  The Psychology Of Intelligence , 1951 .

[2]  Y. Pomeau,et al.  Intermittent transition to turbulence in dissipative dynamical systems , 1980 .

[3]  T. Sejnowski,et al.  Storing covariance with nonlinearly interacting neurons , 1977, Journal of mathematical biology.

[4]  O. Rössler.  The Chaotic Hierarchy , 1983 .

[5]  S. Grossberg.  Neural Networks and Natural Intelligence , 1988 .

[6]  D. J. Wallace,et al.  Models of Neural NetWorks , 1995 .

[7]  G. Bi,et al.  Distributed synaptic modification in neural networks induced by patterned stimulation , 1999, Nature.

[8]  S.-I. Amari,et al.  Neural theory of association and concept-formation , 1977, Biological Cybernetics.

[9]  Shun-ichi Amari,et al.  Statistical neurodynamics of associative memory , 1988, Neural Networks.

[10]  D. Ruelle,et al.  Ergodic theory of chaos and strange attractors , 1985 .

[11]  D. J. Albers,et al.  Routes to Chaos in Neural Networks with Random Weights , 1998 .

[12]  A. Wolf,et al.  Determining Lyapunov exponents from a time series , 1985 .

[13]  Ichiro Tsuda,et al.  Chaotic dynamics of information processing: The “magic number seven plus-minus two” revisited , 1985 .

[14]  Ichiro Tsuda,et al.  Dynamic link of memory--Chaotic memory map in nonequilibrium neural networks , 1992, Neural Networks.

[15]  Daniel J. Amit,et al.  Spike-Driven Synaptic Dynamics Generating Working Memory States , 2003, Neural Computation.

[16]  C. Molter,et al.  Learning cycles brings chaos in Hopfield networks , 2005, Proceedings. 2005 IEEE International Joint Conference on Neural Networks, 2005..

[17]  John F. Kolen,et al.  Understanding and Explaining DRN Behavior , 2001 .

[18]  Hugues Bersini,et al.  Learning Cycles brings Chaos in Continuous Hopfield Networks , 2005 .

[19]  Bruno Cessac,et al.  Self-organization and dynamics reduction in recurrent networks: stimulus presentation and learning , 1998, Neural Networks.

[20]  F. Varela,et al.  Perception's shadow: long-distance synchronization of human brain activity , 1999, Nature.

[21]  K. Kaneko.  Pattern dynamics in spatiotemporal chaos: Pattern selection, diffusion of defect and pattern competition intermettency , 1989 .

[22]  Stefano Fusi,et al.  Hebbian spike-driven synaptic plasticity for learning patterns of mean firing rates , 2002, Biological Cybernetics.

[23]  W. Levy,et al.  Temporal contiguity requirements for long-term associative potentiation/depression in the hippocampus , 1983, Neuroscience.

[24]  A Babloyantz,et al.  Computation with chaos: a paradigm for cortical activity. , 1994, Proceedings of the National Academy of Sciences of the United States of America.

[25]  C. Molter,et al.  Introduction of a Hebbian unsupervised learning algorithm to boost the encoding capacity of Hopfield networks , 2005, Proceedings. 2005 IEEE International Joint Conference on Neural Networks, 2005..

[26]  Ichiro Tsuda,et al.  Chaotic itinerancy , 2013, Scholarpedia.

[27]  A. Grinvald,et al.  Spontaneously emerging cortical representations of visual attributes , 2003, Nature.

[28]  N Brunel,et al.  Slow stochastic Hebbian learning of classes of stimuli in a recurrent neural network. , 1998, Network.

[29]  E. Gardner,et al.  Maximum Storage Capacity in Neural Networks , 1987 .

[30]  Teuvo Kohonen,et al.  Self-organized formation of topologically correct feature maps , 2004, Biological Cybernetics.

[31]  Eric I. Knudsen,et al.  Gated Visual Input to the Central Auditory System , 2002, Science.

[32]  Hugues Bersini,et al.  How chaos in small hopfield networks makes sense of the world , 2003 .

[33]  Hugues Bersini,et al.  The connections between the frustrated chaos and the intermittency chaos in small Hopfield networks , 2002, Neural Networks.

[34]  Kunihiko Kaneko,et al.  ISSUE : Chaotic Itinerancy Chaotic itinerancy , 2003 .

[35]  J. Nicolis.  CHAOTIC DYNAMICS OF INFORMATION PROCESSING WITH RELEVANCE TO COGNITIVE BRAIN FUNCTIONS , 1985 .

[36]  T. Bliss,et al.  Long‐lasting potentiation of synaptic transmission in the dentate area of the anaesthetized rabbit following stimulation of the perforant path , 1973, The Journal of physiology.

[37]  R. Llinás.  I of the Vortex: From Neurons to Self , 2000 .

[38]  I. Tsuda.  Toward an interpretation of dynamic neural activity in terms of chaotic dynamical systems. , 2001, The Behavioral and brain sciences.

[39]  P. Érdi,et al.  The brain as a hermeneutic device. , 1996, Bio Systems.

[40]  E. Gardner,et al.  Three unfinished works on the optimal storage capacity of networks , 1989 .

[41]  John J. Hopfield,et al.  Neural networks and physical systems with emergent collective computational abilities , 1999 .

[42]  Haim Sompolinsky,et al.  Chaos and synchrony in a model of a hypercolumn in visual cortex , 1996, Journal of Computational Neuroscience.

[43]  Nicolas Brunel,et al.  Learning internal representations in an attractor neural network with analogue neurons , 1995 .

[44]  Hugues Bersini,et al.  The frustrated and compositional nature of chaos in small Hopfield networks , 2000, 2000 2nd International Conference. Control of Oscillations and Chaos. Proceedings (Cat. No.00TH8521).

[45]  D. O. Hebb,et al.  The organization of behavior , 1988 .

[46]  W. Freeman,et al.  How brains make chaos in order to make sense of the world , 1987, Behavioral and Brain Sciences.

[47]  H. Sompolinsky,et al.  Chaos in Neuronal Networks with Balanced Excitatory and Inhibitory Activity , 1996, Science.

[48]  E. Rosch,et al.  The Embodied Mind: Cognitive Science and Human Experience , 1993 .

[49]  Walter J. Freeman,et al.  The Hebbian paradigm reintegrated: Local reverberations as internal representations , 1995, Behavioral and Brain Sciences.

[50]  Klaus Schulten,et al.  Neural networks (2nd ed.): an introduction , 1995 .

[51]  Shun-ichi Amari,et al.  Learning Patterns and Pattern Sequences by Self-Organizing Nets of Threshold Elements , 1972, IEEE Transactions on Computers.

[52]  Sommers,et al.  Chaos in random neural networks. , 1988, Physical review letters.

[53]  F. Pasemann.  Complex dynamics and the structure of small neural networks , 2002, Network.

[54]  D. Amit,et al.  Statistical mechanics of neural networks near saturation , 1987 .

[55]  K. Ikeda,et al.  Maxwell-Bloch Turbulence , 1989 .

[56]  Jan-Moritz P Franosch,et al.  How a frog can learn what is where in the dark. , 2005, Physical review letters.

[57]  Eytan Domany,et al.  Models of Neural Networks I , 1991 .

[58]  Daniel J. Amit,et al.  Learning in Neural Networks with Material Synapses , 1994, Neural Computation.