Automated Data Analysis and Discovery in Neurophysiological Simulation Experiments using a Combination of Numerical and Symbolic Methods

Abstract

Raw data derived from experiments or numerical simulations in a certain domain is at an abstraction level far too low to be used for the interpretation by a computer. In order to detect similarities, relations or dependencies within different data sets it is often advisable to construct a qualitative description using various transformational steps. Ideally this process should lead to the same high-level symbolic form that human researchers are used to dealing with themselves. Enabling a system to perform this transformation and to make inferences based on the symbolic description represents a first step towards automatic discovery. In this paper, we report on the above-mentioned aspects of the MOBIS (Modelling of Biological Systems) project, a case-based, interactive simulation environment which is designed to assist neurophysiologists to step through the experiment life-cycle of design, simulation, and analysis of neurophysiological simulation experiments. We are going to present one component of the intelligent assistant whose purpose is to automatically simplify, analyze, and interpret complex numerical neurophysiological data derived from real experiments or — as in our case — from the results of computerized simulation.