Research & Development

Meta Learning is a subfield of machine learning and is often also referred to as “learning to learn”. In Meta Learning, machine learning algorithms are trained to make predictions based on metadata, such as network architecture and data set characteristics, instead of raw data. Therefore, Meta Learning allows for generalization of learning problems and may lead to drastically increased efficiency when substituting otherwise expensive experimentation procedures. At craftworks, we research and use Meta Learning techniques for exactly that.

​

The field of Representation Learning engages with the automated discovery of optimum representations of data given a specific learning problem. Learning the representation instead of spending hours or even days crafting it manually using traditional feature engineering techniques is time saving, often leads to significantly superior outcomes and frequently exposes previously unknown structures in the data. At craftworks, we use Representation Learning techniques to solve challenges ranging from Natural Language Processing to Computer Vision.

​

Applying machine learning to real world problems is not only time consuming and resource intensive, but also challenging. Automated Machine Learning (AutoML) takes care of large parts of the machine learning process, such as feature engineering, model selection and hyperparameter tuning in an autonomous fashion. This frees up time of data scientists and, additionally, enables people not familiar with machine learning to leverage its power. Therefore, a team of our engineers is working on developing cutting-edge AutoML software.

​

Reinforcement Learning is an iterative optimization approach that is motivated by the way humans learn: trial-and-error. In combination with deep learning, Reinforcement Learning is used to teach machines highly complex games, make robots learn arbitrary tasks, optimize control systems of power plants and many more. At craftworks, we recognize the wide area of application and the huge potential of Reinforcement Learning. Therefore, we develop novel approaches for solving multi-agent problems with incomplete information.