|Autoren|| Thomas Grubinger|
|Organisation||Software Competence Center Hagenberg GmbH|
The problem of springback occuring during the process of free form metal sheet bending is described and two possible methods for dealing with springback are discussed. For springback prediction, we present a novel approach to unify regression models learned in parallel on different but related datasets using multi-task feature learning based on symbolic regression. The FFX framework (Fast Function Extraction) is used for symbolic regression. It relies on regularized linear regression instead of genetic programming, thus providing a scalable and deterministic framework for implementation. FFX provides a basis for an iterative multi-task feature learning approach. Models learned on separate tasks are coupled by iteratively promoting common terms and penalizing seldom occurring terms, leading to improved consistency and interpretability of models across tasks and improved stability on new data. The results show that already the use of this basic approach leads to models which share more features, show less complexity and still retain the same model performance when compared to a single symbolic regression run.