Besides their typical classification task, Self-Organizing Maps (SOM) can be used to approximate inputoutput relations. They provide an economic way of storing the essence of past data into input/output support vector pairs. In this paper the SOLIM algorithm (Self-Organising Locally Interpolating Map) is reviewed and an extrapolation method is introduced. This framework allows finding one inver...

The hybridisation of different classification and mining techniques coming from different areas such as the numeric and the symbolic worlds can produce a significant enhancement of the overall classification and retrieval performance in a Data Mining or Information Retrieval context. This paper introduces an experimental methodology to match an explicative structure issued from a symbolic class...

Difficult non-linear problems can be mapped to a set of localised linear problems. A self-organising map (SOM) is used as a gating function to a localised mixture of experts classifier and is shown to find solutions equivalent to those learned by a multi-layer perceptron while retaining the simplicity and resilience of a single-layer perceptron. Modifications to the traditional softmax gate fun...

Understanding and managing water quality in drinking water distribution system is essential for public health and wellbeing, but is challenging due to the number and complexity of interacting physical, chemical and biological processes occurring within vast, deteriorating pipe networks. In this paper we explore the application of Self Organising Map techniques to derive such understanding from ...

When creating datasets for modelling of human skills based on training examples from human motion, one can encounter the problem that the kinematics of the robot does not match the human kinematics. Presented is a simple method of bypassing the explicit modelling of the human kinematics based on a variant of the self-organising map (SOM) algorithm. While the literature contains instances of SOM...

We present the results of a research aimed at improving the Q-learning method through the use of artificial neural networks. Neural implementations are interesting due to their generalisation ability. Two implementations are proposed: one with a competitive multilayer perceptron and the other with a self-organising map. Results obtained on a task of learning an obstacle avoidance behaviour for ...

MOTIVATION Graphical representations of proteins in online databases generally give default views orthogonal to the PDB file coordinate system. These views are often uninformative in terms of protein structure and/or function. Here we discuss the development of a simple automatic algorithm to provide a 'good' view of a protein domain with respect to its structural features. RESULTS We used di...

A fundamental strategy for organising connections in the nervous system is the formation of neural maps. Map formation has been most intensively studied in sensory systems where the central arrangement of axon terminals reflects the distribution of sensory neuron cell bodies in the periphery or the sensory modality. This straightforward link between anatomy and function has facilitated tremendo...

The Self-Organizing Map (SOM), and other related architectures, enjoy a growing popularity in the field of Data Mining. These neural network algorithms provide a topology-preserving mapping from high-dimensional data to a lower dimension, which allows for an easier interpretation of complex data. For visualisation of trained maps, a lot of different techniques have been developed. However, conv...

Statistical re-sampling techniques have been used extensively and successfully in the machine learning approaches for generations of classifier and predictor ensembles. It has been frequently shown that combining so called unstable predictors has a stabilizing effect on and improves the performance of the prediction system generated in this way. In this paper we use the resampling techniques in...