UCL Discovery
UCL home » Library Services » Electronic resources » UCL Discovery

The Graph Cut Kernel for Ranked Data

Conserva, Michelangelo; Deisenroth, Marc Peter; Kumar, KS Sesh; (2022) The Graph Cut Kernel for Ranked Data. Transactions on Machine Learning Research (In press). Green open access

[thumbnail of 2105.12356v2.pdf]
Preview
Text
2105.12356v2.pdf - Published Version

Download (882kB) | Preview

Abstract

Many algorithms for ranked data become computationally intractable as the number of objects grows due to the complex geometric structure induced by rankings. An additional challenge is posed by partial rankings, i.e. rankings in which the preference is only known for a subset of all objects. For these reasons, state-of-the-art methods cannot scale to real-world applications, such as recommender systems. We address this challenge by exploiting the geometric structure of ranked data and additional available information about the objects to derive a kernel for ranking based on the graph cut function. The graph cut kernel combines the efficiency of submodular optimization with the theoretical properties of kernel-based methods. We demonstrate that our novel kernel drastically reduces the computational cost while maintaining the same accuracy as state-of-the-art methods.

Type: Article
Title: The Graph Cut Kernel for Ranked Data
Open access status: An open access version is available from UCL Discovery
Publisher version: https://openreview.net/forum?id=SEUGkraMPi
Language: English
Additional information: © The Authors 2022. Original content in this article is licensed under the terms of the Creative Commons Attribution 4.0 International (CC BY 4.0) Licence (https://creativecommons.org/licenses/by/4.0/)
UCL classification: UCL > Provost and Vice Provost Offices > UCL BEAMS > Faculty of Engineering Science
UCL > Provost and Vice Provost Offices > UCL BEAMS > Faculty of Engineering Science > Dept of Computer Science
UCL > Provost and Vice Provost Offices > UCL BEAMS
UCL
URI: https://discovery.ucl.ac.uk/id/eprint/10153153
Downloads since deposit
45Downloads
Download activity - last month
Download activity - last 12 months
Downloads by country - last 12 months

Archive Staff Only

View Item View Item