Paper presentation at the International Joint Conference on Neural Networks (IJCNN)

25.07.2022 | CAIRO, Pressemeldung
Simon Heilig presented the research paper on Revisiting Memory Efficient Kernel Approximation: An Indefinite Learning Perspective

This week the former FIW student Simon Heilig had the chance to present the work he did together with Prof. Dr. Frank-Michael Schleif and Maximilian Münch to the community of the flagship conference of neural networks from IEEE INNS/CIS. The this year’s IJCNN took place at Padova in collaboration with WCCI 2022.

The work was all about large-scale learning and approximating so-called kernel matrices. The research group demonstrated that a foundamental property was violated by a proposed memory efficient approximation approach. They did not only revisited the method, but they also extended the class of applicable functions, so that arbitrary similarity measures, stemming from real-world scenarios like protein sequence comparisons, can be utilized.
For more details, have a look at the pre-print.

In his LinkedIn post, Heilig expresses "special thanks to the organizers of the conference and to the staff, making this event to a really special starting point in my carrier! Also, I‘m grateful for the support from Center of Artificial Intelligence and Robotics, located at Hochschule für angewandte Wissenschaften Würzburg-Schweinfurt."


Source: https://www.linkedin.com/feed/update/urn:li:activity:6956946344330600448/