maximum entropy

Spoken Language Understanding
slu.pdf in ML/CV

MAP smoothing with Gaussian prior. Chen 99
即原先 expected feature = empirical feature 约束，弱化为
expected feature = empirical feature + v
具体不是很明白