The Mondrian kernel

Conference item

We introduce the Mondrian kernel, a fast random feature approximation to the Laplace kernel. It is suitable for both batch and online learning, and admits a fast kernel-width-selection procedure as the random features can be re-used efficiently for all kernel widths. The features are constructed by sampling trees via a Mondrian process [Roy and Teh, 2009], and we highlight the connection to Mondrian forests [Lakshminarayanan et al., 2014], where trees are also sampled via a Mondrian process, but fit independently. This link provides a new insight into the relationship between kernel methods and random forests.

Authors: Balog, M; Lakshminarayanan, B; Ghahramani, Z; Roy, DM; Teh, YW

• Publication status: Published
• Peer review status: Peer reviewed
• Publisher: AUAI Press
• Host title: UAI'16: Proceedings of the Thirty-Second Conference on Uncertainty in Artificial Intelligence
• Pages: 32-41
• Publication date: 2016-06-25
• Acceptance date: 2016-06-25
• Event title: Thirty-Second Conference on Uncertainty in Artificial Intelligence, UAI 2016
• Event location: New York City, NY, USA
• Event website: https://www.auai.org/uai2016/index.php
• Event start date: 2016-06-25
• Event end date: 2016-06-29
• ISBN: 9780996643115
• Language: English