Train and test tightness of LP relaxations in structured prediction
Name
17-535.pdf
Description
Published version
Size
634.73 KB
Format
Adobe PDF
Checksum (MD5)
4fc90d1870375c2cbad106cb4fc6bec6
Author(s)
Meshi, O
London, B
Weller, A
Sontag, D
Date Issued
February 1, 2019
Journal
Journal of Machine Learning Research
Version
Final published version
Abstract
© 2019 Ofer Meshi, Ben London, Adrian Weller, and David Sontag. Structured prediction is used in areas including computer vision and natural language processing to predict structured outputs such as segmentations or parse trees. In these settings, prediction is performed by MAP inference or, equivalently, by solving an integer linear program. Because of the complex scoring functions required to obtain accurate predictions, both learning and inference typically require the use of approximate solvers. We propose a theoretical explanation for the striking observation that approximations based on linear programming (LP) relaxations are often tight (exact) on real-world instances. In particular, we show that learning with LP relaxed inference encourages integrality of training instances, and that this training tightness generalizes to test data.
MIT Department
Massachusetts Institute of Technology. Computer Science and Artificial Intelligence Laboratory
Terms of Use
Creative Commons Attribution 4.0 International license
Persistent DSpace Link
DOI of Published Version
https://jmlr.org/papers/v20/17-535.html
