The Softmap Algorithm

Authors:
Steven Raekelboom;Marc M. Van Hulle
Affiliations:
Laboratorium voor Neuro- en Psychofysiologie, K.U. Leuven, Campus Gasthuisberg, Herestraat 49, B-3000 Leuven, Belgium. E-mail: Email: marc@neuro,kuleuven.ac.be;Laboratorium voor Neuro- en Psychofysiologie, K.U. Leuven, Campus Gasthuisberg, Herestraat 49, B-3000 Leuven, Belgium. E-mail: Email: marc@neuro,kuleuven.ac.be
Venue:
Neural Processing Letters
Year:
1998

Citing 6
Cited 0

On the stationary state of Kohonen's self-organizing sensory mapping

Biological Cybernetics
On an unsupervised learning rule for scalar quantization following the maximum entropy principle

Neural Computation
Self-organization as an iterative kernel smoothing process

Neural Computation
Topology-preserving map formation achieved with a purely local unsupervised competitive learning rule

Neural Networks
Self-organizing maps

Self-organizing maps
Topographic map formation by maximizing unconditional entropy: a plausible strategy for “online” unsupervised competitive learning and nonparametric density estimation

IEEE Transactions on Neural Networks

Quantified Score

Hi-index	0.00

Visualization

Abstract

A new unsupervised competitive learning rule is introduced, called theSelf-organizing free-topology map (Softmap) algorithm, fornonparametric density estimation. The receptive fields of the formalneurons are overlapping, radially-symmetric kernels, the radii of whichare adapted to the local input density together with the weight vectorswhich define the kernel centers. A fuzzy code membership function isintroduced in order to encompass, in a novel way, the presence ofoverlapping receptive fields in the competitive learning scheme.Furthermore, a computationally simple heuristic is introduced fordetermining the overall degree of smoothness of the resulting densityestimate. Finally, the density estimation performance is compared tothat of the variable kernel method, VBAR and Kohonen‘s SOM algorithm.