We present a pairwise learning to rank approach based on a neural net, called DirectRanker, that generalizes the RankNet architecture. We show mathematically that our model is reflexive, antisymmetric, and transitive allowing for simplified training and improved performance. Experimental results on the LETOR MSLR-WEB10K, MQ2007 and MQ2008 datasets show that our model outperforms numerous state-of-the-art methods, while being inherently simpler in structure and using a pairwise approach only.
|Title of host publication||Machine Learning and Knowledge Discovery in Databases - European Conference, ECML PKDD 2019, Proceedings|
|Editors||Ulf Brefeld, Elisa Fromont, Andreas Hotho, Arno Knobbe, Marloes Maathuis, Céline Robardet|
|Number of pages||16|
|Publication status||Published - 2020|
|Event||European Conference on Machine Learning and Principles and Practice of Knowledge Discovery in Databases, ECML PKDD 2019 - Wurzburg, Germany|
Duration: 16 Sep 2019 → 20 Sep 2019
|Name||Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)|
|Conference||European Conference on Machine Learning and Principles and Practice of Knowledge Discovery in Databases, ECML PKDD 2019|
|Period||16/09/19 → 20/09/19|
Bibliographical noteFunding Information:
We would like to thank Dr. Christian Schmitt for his contributions to the work presented in this paper. We also thank Luiz Frederic Wagner for proof(read)ing the mathematical aspects of our model. Parts of this research were conducted using the supercomputer Mogon and/or advisory services offered by Johannes Gutenberg University Mainz (hpc.uni-mainz.de), which is a member of the AHRP (Alliance for High Performance Computing in Rhineland Palatinate, www.ahrp.info) and the Gauss Alliance e.V. The authors gratefully acknowledge the computing time granted on the supercomputer Mogon at Johannes Gutenberg University Mainz (hpc.uni-mainz.de). This research was partially funded by the Carl Zeiss Foundation Project: ?Compe-tence Centre for High-Performance-Computing in the Natural Sciences? at the University of Mainz. Furthermore, Andreas Karwath has been co-funded by the MRC grant MR/S003991/1.
This research was partially funded by the Carl Zeiss Foundation Project: ‘Competence Centre for High-Performance-Computing in the Natural Sciences’ at the University of Mainz. Furthermore, Andreas Karwath has been co-funded by the MRC grant MR/S003991/1.
© Springer Nature Switzerland AG 2020.
- Information retrieval
- Learning to rank
- Machine learning
ASJC Scopus subject areas
- Theoretical Computer Science
- Computer Science(all)