CS229 Project Report: NER adaptation
msra(2005)
摘要
For each transition probability, MEMM uses maximum entropy model. Maximum entropy model is used to model the data distribution, which gives us as much information as possible. Without any constraints, the uniform distribution gives us maximum entropy. However, we have training data which gives some facts about the true distribution. What maximum entropy model does is to maximize the entropy of the data give such constraints coming from the training data. The constraints are that the expected value of each feature in the distribution is the same as its actual count in the training data. I.e.
更多查看译文
AI 理解论文
溯源树
样例
生成溯源树,研究论文发展脉络