Embedding Inference for Structured Multilabel Prediction
- Farzaneh Mirzazadeh, University of Alberta
- Siamak Ravanbakhsh, Dept of Computing Science, University of Alberta
- Nan Ding
- Dale Schuurmans, AICML
A key bottleneck in structured output prediction is the need for inference during training and testing, usually requiring some form of dynamic programming. Rather than using approximate inference or tailoring a specialized inference method for a particular structure---standard responses to the scaling challenge---we propose to embed prediction constraints directly into the learned representation. By eliminating the need for explicit inference a more scalable approach to structured output prediction can be achieved, particularly at test time. We demonstrate the idea for multi-label prediction under subsumption and mutual exclusion constraints, where a relationship to maximum margin structured output prediction can be established. Experiments demonstrate that the benefits of structured output training can still be realized even after inference has been eliminated.
Citation
F. Mirzazadeh, S. Ravanbakhsh, N. Ding, D. Schuurmans. "Embedding Inference for Structured Multilabel Prediction". Neural Information Processing Systems (NIPS), (ed: C. Cortes, N. D. Lawrence, D. D. Lee, M. Sugiyama, R. Garnett), pp 3555-3563, December 2015.| Keywords: | |
| Category: | In Conference |
| Web Links: | NeurIPS |
BibTeX
@incollection{Mirzazadeh+al:NIPS15,
author = {Farzaneh Mirzazadeh and Siamak Ravanbakhsh and Nan Ding and Dale
Schuurmans},
title = {Embedding Inference for Structured Multilabel Prediction},
Editor = {C. Cortes, N. D. Lawrence, D. D. Lee, M. Sugiyama, R. Garnett},
Pages = {3555-3563},
booktitle = {Neural Information Processing Systems (NIPS)},
year = 2015,
}Last Updated: February 14, 2020Submitted by Sabina P
