Advances in Self-Organising Maps

This is the third Workshop on Self-Organising Maps (WSOM) and its related techniques. The previous two were held in Helsinki (1997 and 1999) and confIrmed the vitality of the SOM as one of the most popular and powerful concepts for unsupervised pattern recognition and data visualisation. These meetings not only acted as a showcase for the latest advances in SOM theory and for illustrating its vast range of applicability, but also as venues where much informal and fruitful interaction could take place. It is interesting to observe the development of the original SOM, and this remarkable progress confrrms the originality and insight of Teuvo Kohonen's pioneering work. With the range and quality of the papers in this volume, the stage is set for another very successful meeting. This volume is a permanent record of all the contributions presented during WSOM'OI held at the University of Lincolnshire and Humberside, 13 - 15 June, 2001. The University is the newest of England's universities but it is situated in the heart of one of our oldest cities - founded by the Romans and overlooked by the towering mass of its medieval cathedral. Primarily Lincoln has always been a centre for the rich agricultural heartland of England; however, it was the birthplace, 186 years ago, of George Boole. So WSOM'OI is continuing Lincoln's long and honourable tradition of advancing scientific understanding.

Towards an information-theoretic approach to kernel-based topographic map formation.-A statistical tool to assess the reliability of self-organising maps.-A SOM association network.-A supervised self-organising map for structured data.-Human gait analysis using SOM.-Exploring financial crises data with self-organising maps (SOM).-Analysing health inequalities using SOM.-Integrating contextual information into text document clustering with self-organising maps.-Self-organising Internet semantic network.-Recursive learning rules for SOMs.-Induced Vorono kernels for principal manifolds approximation.-Visualisation induced SOM (ViSOM).-An approach to automated interpretation of SOM.-VQ-based clustering algorithm of piecewise-dependant-data.-Adaptive subspace encoders using stochastic vector quantisers.-SHAPESOM.-A new interpolation algorithm employing a self organising map.-SOM-based exploratory analysis of gene expression data.-Exploring power transformer database using self-organising maps (SOM) and minimal spanning tree (MST).-Recent advances with the growing hierarchical self-organising Map.-Self-organising maps of web link information.-A design method of DNA chips using self-organising maps.-Multi-dimensional self-organising maps on massively parallel hardware.-A new method of Hough transform by using SOM with input vector transformation.-Estimating relevant input dimensions for self-organising algorithms.- An overview on unsupervised learning from data mining perspective.-Recursive self-organising maps.-An investigation into catastrophic interference on a SOM network.-A topography-preserving latent variable model with learning metrics.-Vector quantisation with g-observable neighbors.-Dynamic vector quantisation of speech.-Optimisation of electronic parts mounting machines using SOM-TSP method with 5 dimensional data.-Self-organising maps forcondition assessment of paper insulated cables.-An essay in classifying self-organising maps for temporal sequence processing.-Evaluating SOM-based models in text classification tasks for the Greek language.-Nonlinear blind source separation using SOMs and applications to image denoising.-Signal-based feature extraction and SOM based dimension reduction in a vibration monitoring microsystem.