Conditional Random Fields: Probabilistic Models for Segmenting and Labeling Sequence Data
ICML '01 Proceedings of the Eighteenth International Conference on Machine Learning
Support vector machine learning for interdependent and structured output spaces
ICML '04 Proceedings of the twenty-first international conference on Machine learning
Training conditional random fields via gradient tree boosting
ICML '04 Proceedings of the twenty-first international conference on Machine learning
Fast inference and learning in large-state-space HMMs
ICML '05 Proceedings of the 22nd international conference on Machine learning
EMNLP '02 Proceedings of the ACL-02 conference on Empirical methods in natural language processing - Volume 10
Incremental parsing with the perceptron algorithm
ACL '04 Proceedings of the 42nd Annual Meeting on Association for Computational Linguistics
Scaling conditional random fields using error-correcting codes
ACL '05 Proceedings of the 43rd Annual Meeting on Association for Computational Linguistics
Solving multiclass learning problems via error-correcting output codes
Journal of Artificial Intelligence Research
Sensitive error correcting output codes
COLT'05 Proceedings of the 18th annual conference on Learning Theory
| Hi-index | 0.00 |
Markov models are commonly used for joint inference of label sequences. Unfortunately, inference scales quadratically in the number of labels, which is problematic for training methods where inference is repeatedly preformed and is the primary computational bottleneck for large label sets. Recent work has used output coding to address this issue by converting a problem with many labels to a set of problems with binary labels. Models were independently trained for each binary problem, at a much reduced computational cost, and then combined for joint inference over the original labels. Here we revisit this idea and show through experiments on synthetic and benchmark data sets that the approach can perform poorly when it is critical to explicitly capture the Markovian transition structure of the large-label problem. We then describe a simple cascade-training approach and show that it can improve performance on such problems with negligible computational overhead.