Algorithmic intelligence as an emergent phenomenon
Machine learning has made tremendous progress in the last years. Recent breakthroughs in deep learning give computers new access to complex real-world data, with far-reaching consequences for industry, science and society. The purpose of this project is to review the structural basis of (algorithmic) intelligence in order to better understand the limitations and possibilities of known learning methods and to contribute to the development of new, more powerful methods.
Central to this approach is the hypothesis that (natural) intelligence has evolved through adaptation to basic structural features of the natural world. As a result, algorithmic developments in computer science can be better understood when we interpret them from the perspective of empirical science - in particular: statistical physics and evolutionary biology. In the established research centre methods of statistical learning will be considered from an interdisciplinary perspective. The research centre supports interdisciplinary projects in which classical empirical sciences (physics, biology) and information sciences (mathematics, computer science) work together to better understand existing methods of machine learning and/or to develop new methods.
The funding for the interdisciplinary research centre in the Natural Sciences of Johannes Gutenberg University Mainz is provided by the Carl Zeiss Foundation.