31.12.2030
| DSC Seed Grant |
The DSC supports a research project on methods for FDR control in ML applications
As part of this project, the DSC Seed Grant funds a research stay in Tel Aviv, Israel. The international collaboration aims to further develop knock-off methods for false discovery rate (FDR) control in novelty detection applications.
The project is a continuation of a research collaboration between Pascal Rink (University of Bremen), Prof. Ruth Heller (Tel Aviv University - TAU) and her PhD student Tzviel Frostig which started earlier this year in Israel. During research on existing knock-off methods for controlling FDR in novelty detection applications, the method AdaDetect turned out to be particularly promising.
The method AdaDetect is used for novelty detection. Novelty detection is a process that is employed to check whether certain observations lie within or outside of a given distribution. AdaDetect combines existing methods for FDR control and machine learning (ML) applications. It is able to control the FDR of ML classifiers in high-dimensional applications. Moreover, AdaDetect has great power properties and can be applied to any type of classification algorithm, including, for instance, random forest or neural networks.
The research stay in Tel Aviv aims to further develop AdaDetect by adjusting it to specific ML applications and studying the impact on quality. AdaDetect is one of the first works that brings together FDR control and machine learning. Thus, the method has application potential is several areas. These include, for example, the detection of stars in astrophysics or the identification of gene features in certain disease patterns in the field of biostatistics.
Furthermore, the joint project intends to encourage the exchange between statistical researchers from both universities and thereby pave the way for more international collaborations in the future.
Funding recipient:
Pascal Rink (Faculty 03 – Mathematics and Computer Science)
Funding period:
07 September 2023 - 09 November 2023
Updated by: Svenja Goers
« back
31.12.2030 | DSC Seed Grant
The DSC supports a research project on methods for FDR control in ML applications
As part of this project, the DSC Seed Grant funds a research stay in Tel Aviv, Israel. The international collaboration aims to further develop knock-off methods for false discovery rate (FDR) control in novelty detection applications.
The project is a continuation of a research collaboration between Pascal Rink (University of Bremen), Prof. Ruth Heller (Tel Aviv University - TAU) and her PhD student Tzviel Frostig which started earlier this year in Israel. During research on existing knock-off methods for controlling FDR in novelty detection applications, the method AdaDetect turned out to be particularly promising.
The method AdaDetect is used for novelty detection. Novelty detection is a process that is employed to check whether certain observations lie within or outside of a given distribution. AdaDetect combines existing methods for FDR control and machine learning (ML) applications. It is able to control the FDR of ML classifiers in high-dimensional applications. Moreover, AdaDetect has great power properties and can be applied to any type of classification algorithm, including, for instance, random forest or neural networks.
The research stay in Tel Aviv aims to further develop AdaDetect by adjusting it to specific ML applications and studying the impact on quality. AdaDetect is one of the first works that brings together FDR control and machine learning. Thus, the method has application potential is several areas. These include, for example, the detection of stars in astrophysics or the identification of gene features in certain disease patterns in the field of biostatistics.
Furthermore, the joint project intends to encourage the exchange between statistical researchers from both universities and thereby pave the way for more international collaborations in the future.
Funding recipient:
Pascal Rink (Faculty 03 – Mathematics and Computer Science)
Funding period:
07 September 2023 - 09 November 2023
Author: Svenja Goers
If you have any questions regarding the DSC Seed Grant, feel free to contact me.
Dr. Lena Steinmann
DSC Coordinator
+49 (421) 218 - 63941
lena.steinmann@uni-bremen.de
« back