Do Deep Generative Models Know What They Don't Know?, Eric Nalisnick, Akihiro Matsukawa, Yee Whye Teh, and Balaji Lakshminarayanan, 2018International Conference on Learning Representations (ICLR) (International Conference on Learning Representations (ICLR))DOI: 10.48550/arXiv.1802.04013 - 这篇论文强调了使用深度生成模型(包括VAE)的原始似然进行可靠的域外检测的局限性,为基于VAE的异常评分提供重要背景。