Some studies of intrinsic dimensionality with interactive graphics

In pattern recognition and other forms of data analysis, the investigator would like to find the parameters which most efficiently describe the data. This is useful for feature extraction and to gain insight into the meaning of the data. A number of techniques have been developed to estimate the intrinsic dimensionality of a data set. This paper describes a technique which utilizes the interactive capabilities of computer graphics to find a nonlinear transformation to project the data into its intrinsic parameter space.