|Current Workshop | Overview||Back | Print | Home | Search ||
Lorentz Center Highlights
Facticity, Creativity and Complexity
University of Amsterdam
ABSTRACT: Science in the 21st century has to deal with increasingly large and complex data sets. This calls for innovative techniques for the analysis and description of these data sets. At our group in Amsterdam we have been working on measuring complexity of large date sets using Kolmogorov complexity as a theoretical foundation. The theory of facticity tries to find a fundamental answer to the following question: what can I learn from a data set? Or, to put it differently: how much model information does a data set contain. It does this by studying the balance between ad hoc and structural data under (optimal) compression. Some of our findings are not really surprising (there is no real optimal model), others are: creative systems (that constantly increase their model complexity) will eventually collapse. Extremely large graph based knowledge bases, such as the Internet, also seem to call for new compression techniques. I will present some of these ideas and discuss some consequences.