Prof. Dr. Jürgen Döllner


"Customizable Asymmetric Loss Functions for Machine Learning-based Predictive Maintenance" accepted for Condition Monitoring and Maintenance 2020

An asymmetric loss function with exponential growth on both sides in contrast to the symmetric Mean Squared Error loss.

Paper accepted for CMD2020

A paper titled "Customizable Asymmetric Loss Functions for Machine Learning-based Predictive Maintenance" by Lukas Ehrig, Daniel Atzberger, Benjamin Hagedorn, Jan Klimke, and Jürgen Döllner was accepted for presentation and publication for the 8th International Conference on Condition Monitoring and Diagnosis (CMD 2020) .

The paper introduces an approach of modelling different costs for overestimation and underestimation within machine learning approaches for predicitve maintenance. It describes a configureable cost function that is can be used for training of machine learning techniques in order to optimize predictions that are made using these models.

Due to the current situation concerning the COVID-19 pandemic, the conference will take place as a virtual conference from October 25th to October 28th 2020.


This project has been partially funded by:

  • Federal Ministry of Economic Affairs and Energy under ZIM Program (Project IBDAV)
  • German Federal Ministry of Education and Research BMBF: AI Lab for IT-Systems Engineering