Where the abstract feature maps of the brain might come from.

Three types of neuronal organization can be called 'brain maps': sets of feature-sensitive cells, ordered projections between neuronal layers and ordered maps of abstract features. The latter are most intriguing as they reflect the central properties of an organism's experiences and environment. It is proposed that such feature maps are learned in a process that involves parallel input to neurons in a brain area and adaptation of neurons in the neighborhood of the cells that respond most strongly to this input. This article presents a new mathematical formulation for such adaptation and relates it to physiological functions.

Variants of self-organizing maps.

Self-organizing maps have a bearing on traditional vector quantization. A characteristic that makes them more closely resemble certain biological brain maps, however, is the spatial order of their responses, which is formed in the learning process. A discussion is presented of the basic algorithms and two innovations: dynamic weighting of the input signals at each input of each cell, which improves the ordering when very different input signals are used, and definition of neighborhoods in the learning algorithm by the minimal spanning tree, which provides a far better and faster approximation of prominently structured density functions. It is cautioned that if the maps are used for pattern recognition and decision process, it is necessary to fine tune the reference vectors so that they directly define the decision borders.

Self organization of a massive document collection.

This article describes the implementation of a system that is able to organize vast document collections according to textual similarities. It is based on the self-organizing map (SOM) algorithm. As the feature vectors for the documents statistical representations of their vocabularies are used. The main goal in our work has been to scale up the SOM algorithm to be able to deal with large amounts of high-dimensional data. In a practical experiment we mapped 6,840,568 patent abstracts onto a 1,002,240-node SOM. As the feature vectors we used 500-dimensional vectors of stochastic figures obtained as random projections of weighted word histograms.

Adaptive, associative, and self-organizing functions in neural computing.

This paper contains an attempt to describe certain adaptive and cooperative functions encountered in neural networks. The approach is a compromise between biological accuracy and mathematical clarity. two types of differential equation seem to describe the basic effects underlying the information of these functions: the equation for the electrical activity of the neuron and the adaptation equation that describes changes in its input connectivities. Various phenomena and operations are derivable from them: clustering of activity in a laterally interconnected nework; adaptive formation of feature detectors; the autoassociative memory function; and self-organized formation of ordered sensory maps. The discussion tends to reason what functions are readily amenable to analytical modeling and which phenomena seem to ensue from the more complex interactions that take place in the brain.

Comparison of SOM point densities based on different criteria.

Point densities of model (codebook) vectors in self-organizing maps (SOMs) are evaluated in this article. For a few one-dimensional SOMs with finite grid lengths and a given probability density function of the input, the numerically exact point densities have been computed. The point density derived from the SOM algorithm turned out to be different from that minimizing the SOM distortion measure, showing that the model vectors produced by the basic SOM algorithm in general do not exactly coincide with the optimum of the distortion measure. A new computing technique based on the calculus of variations has been introduced. It was applied to the computation of point densities derived from the distortion measure for both the classical vector quantization and the SOM with general but equal dimensionality of the input vectors and the grid, respectively. The power laws in the continuum limit obtained in these cases were found to be identical.

Self-organized formation of colour maps in a model cortex.

A report is presented of computer simulations which demonstrate the applicability of self-organization principles to the formation of a cortical colour map. A two-dimensional array of cortical units can be shown to become selectively sensitive to different colour stimuli in an orderly fashion. The precortical part of the simulation model contains three wavelength-sensitive receptors with overlapping sensitivity distributions, and a simple opponent processing stage. Each cortical unit receives the same activity from the precortical stage by adaptive connections. Initially the connections are arbitrary; during self-organization they are changed so that different way. Self-organization of the connections can be shown to take place even when the stimulation of the receptors is totally random, ie the wavelength and purity (and in some simulations the distribution) of each stimulus are selected randomly.

Associative memory and pattern recognition.

This tutorial review presents a model of neural associative memory along with a set of computer demonstrations showing the relevance of this memory mechanism in visual information processing. The model is based upon the hypothesis that adaptive changes in neural networks are intermediated by changes in synaptic efficacies. The signal patterns are stored by gradual changes of the network and they may be recalled later using a part of the original signal pattern as a key. The ability of this type of memory mechanism to process sensory information is emphasized.