Data visualisation using neural networks

The problem of analysing a dataset which has many variables is one common to many disciplines, from sociology to molecular chemistry. Many techniques are employed in an attempt to detect patterns or trends within the datasets. RENDER is a neural net based algorithm for mapping high-dimensional dataset into two dimensions, so that the human eye can be used to detect trends and patterns in the data. It combines the advantages of linear and nonlinear mapping, in that it is relatively inexpensive to compute, it is reversible and does not lose as much information as a linear method. However the mathematics are not as precise as either of the other two methods, which may lead to difficulty with interpreting results. Further developments are in hand, both to improve the algorithm and its mathematical basis, and to develop it as a software tool