Conditional Random Fields Training System
(Redirected from CRF toolkit)
Jump to navigation
Jump to search
A Conditional Random Fields Training System is a model-based learning system that can solve a CRF training task (by applying a CRF algorithms).
- Context:
- It can range from being a CRF Library to being a CRF Toolkit to being a CRF Package.
- …
- Example(s):
- MALLET Toolkit.
- iitb.CRF Package.
- CRFpp Package.
- Kevin Murphy CRF Toolbox.
- PyStruct[1] “PyStruct actually implements perceptron and max-margin learning, not maximum likelihood learning. So these models might better be called Maximum Margin Random Fields. However, in the computer vision community, it seems most pairwise models are called CRFs, independent of the method of training.”
- Counter-Example(s):
- a Factor Graph Toolkit, like FACTORIE.
- an HMM Toolkit, like ...
- an SVM Toolkit, like SVMlight.
- a Decision Tree Toolkit, like C4.5.
- a Logistic Regression Toolkit, like LIBLINEAR.
- See: Data Mining Package.
References
2010
- (Cheng, Sun, et al., 2010) ⇒ Yong Cheng, Chengjie Sun, Lei Lin, Yuanchao Liu. (2010). “A Comparison Study of Conditional Random Fields Toolkits.” In: Proceedings of the 6th International Conference on Intelligent Computing, (ICIC 2010). doi:10.1007/978-3-642-14831-6_26
- ABSTRACT: Conditional random fields (CRF) model is an important and widely used sequence labeling model. In this paper, we introduce several commonly used CRF toolkits. Through the analysis and comparison of the toolkits, we give each one’s advantages and disadvantages. We also count the popularity and applicable fields of them. At last, we give our comments for each toolkit.