PUB - Publikationen an der Universität Bielefeld

This paper introduces Mode Explorer, a novel interactive auditory data exploration method to investigate features of high-dimensional data distributions: scratching-interactions on a 2D scatter plot of high-dimensional data with a pencil induces real-time dynamical processes according to a particle sonification model, excited in data space at the nearest mode in the probability density function (pdf) obtained by Kernel Density Estimation. Specifically, the sign-inverted pdf is used as potential energy function in which test particles perform oscillations at low friction, yielding trajectories, and via their instantaneous kinetic energy signals that are directly played back as sound. This Model-based sonification approach enables an interactive search for different modes, to investigate their details, e.g., comparing cluster mass. We present results of a user study, which allows us to conclude that the Mode Explorer enhances users' ability to discriminate clusters and to compare their relative a-priori probabilities.
**V1.mov**: Demonstration of sniffing the particles as they traveled through the data space and converged to the mode with a low friction. **V2.mov**: Demonstration video of the Mode Explorer using a Wacom Tablet and modified stylus. **trajectory.txt**: Cython code of the potential function. Note: This code only demonstrates the algorithm, to use it Cython library needs to be installed. We will release a full demonstration program for public use for the camera ready version.