Locality of global stochastic interaction in directed acyclic networks

Authors:
Nihat Ay
Affiliations:
Max-Planck-Institute for Mathematics in the Sciences, 04103 Leipzig, Germany
Venue:
Neural Computation
Year:
2002

Citing 10
Cited 6

Self-Organization in a Perceptual Network

Computer
Elements of information theory

Elements of information theory
An information-maximization approach to blind separation and blind deconvolution

Neural Computation
A local learning rule that enables information maximization for arbitrary input distributions

Neural Computation
Natural gradient works efficiently in learning

Neural Computation
Information maximization and independent component analysis: is there a difference?

Neural Computation
Spikes: exploring the neural code

Spikes: exploring the neural code
Exploring randomness

Exploring randomness
Probabilistic Networks and Expert Systems

Probabilistic Networks and Expert Systems
Information geometry on hierarchy of probability distributions

IEEE Transactions on Information Theory

Dynamical properties of strongly interacting Markov chains

Neural Networks
Finite State Automata Resulting from Temporal Information Maximization and a Temporal Learning Rule

Neural Computation
Representations of Space and Time in the Maximization of Information Flow in the Perception-Action Loop

Neural Computation
Recurrent infomax generates cell assemblies, neuronal avalanches, and simple cell-like selectivity

Neural Computation
Stochastic interaction in associative nets

Neurocomputing
A temporal learning rule in recurrent systems supports high spatio-temporal stochastic interactions

Neurocomputing

Quantified Score

Hi-index	0.00

Visualization

Abstract

The hypothesis of invariant maximization of interaction (IMI) is formulated within the setting of random fields. According to this hypothesis, learning processes maximize the stochastic interaction of the neurons subject to constraints. We consider the extrinsic constraint in terms of a fixed input distribution on the periphery of the network. Our main intrinsic constraint is given by a directed acyclic network structure. First mathematical results about the strong relation of the local information flow and the global interaction are stated in order to investigate the possibility of controlling IMI optimization in a completely local way. Furthermore, we discuss some relations of this approach to the optimization according to Linsker's Infomax principle.