KI-Lab – Trading off Non-Functional Properties of Machine learning

KI-Lab – Trading off Non-Functional Properties of Machine learning i3mainz, CC BY SA 4.0

How decentralized should data be stored to protect privacy, and how does that affect energy consumption? Conflicting goals of this kind are analyzed in the TOPML project (Trading Off Non-Functional Properties of Machine Learning) Research Center for Machine Learning. In the AI lab at Mainz University of Applied Sciences, the findings from the research center are applied to industrial practice.


The KI-Lab is part of an interdisciplinary research center for machine learning at the Johannes Gutenberg University and the [University of Mainz](https://www.hs-mainz. de/). Here, interactions and dependencies of various properties of machine learning are to be analyzed and weighed.

The transparency and fairness of data and algorithms as well as data protection and the efficient use of resources such as electricity are examined. The focus is on competing needs:

  • How decentralized can data be stored and processed to protect privacy?
  • To what extent does this affect the transparency of algorithms and data?
  • What impact does this have on energy consumption?

The various trade-offs are identified and characterized in order to create viable trade-offs for the application.


Previous activities in the context of the AI ​​lab:

The GDV Society for Geographical Data Processing mbH is building an automated control for EU subsidies in the agricultural sector. Can transparency in machine learning also contribute to resource conservation here?

The BD-A is a specialist in predicting the needs of a product from past experience. What are the disadvantages and advantages of transparent and explainable models compared to black-box models of machine learning?