Tags

今天看了几篇文章,大体上对noise-contrastive estimation 有了些理解,文献主要有:
[1] Noise-contrastive estimation of unnormalized statistical models, with applications to natural image statistics;
[2] Noise-contrastive estimation: A new estimation principle for unnormalized statistical models;
[3] Generalized Noise Contrastive Estimation Motivation  这是一个PPT
[4] Domain Adaptation for Sequence Labeling Tasks with a Probabilistic Language Adaptation Model 这是一篇应用noise-contrastive estimation 解决domain adaptation POS tagger的文章.

总体来说,noise-contrastive estimation 的目的是对一群observed data的概率密度进行估计的方法.通过构造一些分布已知的noise data,将概率密度的估计问题转化为一个logistic regression问题.

拿文献[1]来说,设待估计概率密度p_t