arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 1901.00612cs.LGcs.AIstat.ML
收藏

基于非规范化分布的采样器对抗学习

Adversarial Learning of a Sampler Based on an Unnormalized Distribution

Microsoft Research(微软研究院);Duke University(杜克大学);University at Buffalo(布法罗大学)

Chunyuan Li, Ke Bai, Jianqiao Li, Guoyin Wang, Changyou Chen, Lawrence Carin

摘要:

我们研究了只有非标准化形式的密度而不是样本可以访问的情况下的对抗学习。有了这样的见解,对抗学习被扩展到可以访问目标密度函数的非归一化形式u(x),但没有样本的情况。此外,GAN正则化中的新概念是基于从样本或从u(x)学习而开发的。将所提出的方法与其他方法进行比较,在一系列应用中展示了令人鼓舞的结果,包括深度软Q学习。

英文摘要:

We investigate adversarial learning in the case when only an unnormalized form of the density can be accessed, rather than samples. With insights so garnered, adversarial learning is extended to the case for which one has access to an unnormalized form u(x) of the target density function, but no samples. Further, new concepts in GAN regularization are developed, based on learning from samples or from u(x). The proposed method is compared to alternative approaches, with encouraging results demonstrated across a range of applications, including deep soft Q-learning.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑