CRF Model Parameter Estimation Task

A CRF Model Parameter Estimation Task is a graphical model training task that can produce a CRF-based classifier.

• AKA: CRF Training.
• Context:
  • It can be solved by a CRF Training System (that implements a CRF training algorithm).
• Counter-Example(s):
  • an HMM Training Task.
  • a Decision Tree Training Task.
  • a CRF Inference Task.
• See: CRF Label Bias, CRF Length Bias.

---

---

References

2010

• (Lavergne et al., 2010) ⇒ Thomas Lavergne, Olivier Cappé, and François Yvon. (2010). “Practical Very Large Scale CRFs.” In: Proceedings of the 48th Annual Meeting of the Association for Computational Linguistics, (ACL 2010).
  • QUOTE: In this paper, we address the issue of training very large CRFs, containing up to hundreds output labels and several billion features. Efficiency stems here from the sparsity induced by the use of a l¹ penalty term. Based on our own implementation, we compare three recent proposals for implementing this regularization strategy.

2006

• (Vishwanathan et al., 2006) ⇒ S. V. N. Vishwanathan, Nicol N. Schraudolph, Mark W. Schmidt, and Kevin P. Murphy. (2006). “Accelerated Training of Conditional Random Fields with Stochastic Gradient Methods.” In: Proceedings of the 23rd International Conference on Machine learning (ICML-2006). doi:10.1145/1143844.1143966
  • QUOTE: We apply Stochastic Meta-Descent (SMD), a stochastic gradient optimization method with gain vector adaptation, to the training of Conditional Random Fields (CRFs).

---