Generalization Bounds for Ranking Algorithm via Query-Level Stabilities Analysis
The effectiveness of ranking algorithms determine the quality of information retrieval and the goal of ranking algorithms are to learn a real-valued ranking function that induces a ranking or ordering over an instance space. We focused on generalization ability of learning to rank algorithms for information retrieval (IR). As a continuous research of generalization bounds of ranking algorithm, the contribution of this paper includes: generalization bounds for such ranking algorithm via five kinds of stabilities were given. Such stabilities have lower demand than uniform stability and fit for more real applications.
Keywordsranking algorithmic stability generalization bounds strong stability weak stability
Unable to display preview. Download preview PDF.
- 1.He, X., Gao, W., Jia, Z.: Generalization bounds of Ranking via Query-Level Stability. In: Proceedings of 2011 2nd International Conference on Intelligent Transportation Systems and Intelligent Computing (ITSIC 2011), Suzhou, China (June 2011)Google Scholar
- 2.Lan, Y., Liu, T., Qin, T., Ma, Z., Li, H.: Query-Level Stability and Generalization in Learning to Rank. In: Appearing in Proceedings of the 25 th International Conference on Machine Learning, Helsinki, Finland (2008)Google Scholar
- 3.McDiarmid, C.: On the method of bounded differences. In: Surveys in Combinatorics 1989, pp. 148–188. Cambridge University Press, Cambridge (1989)Google Scholar
- 4.Kutin, S.: Extensions to McDiarmid’s inequality when differences are bounded with high probability, Technical report, Department of Computer Science, The university of Chicago (2002)Google Scholar