Auto-Tune: Automatic structure optimization of learning algorithms on big datasets

Grant BR 3815/8-1 of the Priority Program 1527 Autonomous Learning

Project members

Dr. Frank Hutter
Dr. Philipp Hennig (MPI for Intelligent Systems)
Prof. Dr. Thomas Brox
Aaron Klein

Abstract

We aim to automate the design of machine learning algorithms, in order to facilitate their use by non-experts. Although automated adaptation is a core idea of machine learning, most algorithms still require a choice of external design parameters by an expert, which limits their commercial success. Finding the best instantiation of an existing learning framework is an optimization problem: search through the combined space of different machine learning algorithms, and find the best one. Because this space of possible setups is very large, we will use assumptions and prior knowledge about regularity among, and interdependence between different setups to guide the search efficiently.