arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~
arXiv 1902.02767cs.LGstat.ML
收藏

具有深层可逆特征的混合模型

Hybrid Models with Deep and Invertible Features

DeepMind(深度思维)

Eric Nalisnick, Akihiro Matsukawa, Yee Whye Teh, Dilan Gorur, Balaji Lakshminarayanan

摘要:

我们提出了一种神经混合模型,该模型由一组通过深度可逆变换(即归一化流)计算的特征上定义的线性模型组成。我们的模型的一个有吸引力的性质是,p(特征),特征的密度,和p(目标)|特征),即预测分布,可以在单个前馈过程中精确计算。我们表明,我们的混合模型,尽管可逆性的限制,实现了类似的准确性,纯粹的预测模型。此外,尽管混合优化目标,生成组件仍然是输入特征的良好模型。这提供了额外的功能,例如检测分布外输入和启用半监督学习。精确的联合密度p(目标,特征)的可用性也使我们能够轻松计算许多数量,使我们的混合模型成为概率深度学习下游应用的有用构建块。

英文摘要:

We propose a neural hybrid model consisting of a linear model defined on a set of features computed by a deep, invertible transformation (i.e. a normalizing flow). An attractive property of our model is that both p(features), the density of the features, and p(targets | features), the predictive distribution, can be computed exactly in a single feed-forward pass. We show that our hybrid model, despite the invertibility constraints, achieves similar accuracy to purely predictive models. Moreover the generative component remains a good model of the input features despite the hybrid optimization objective. This offers additional capabilities such as detection of out-of-distribution inputs and enabling semi-supervised learning. The availability of the exact joint density p(targets, features) also allows us to compute many quantities readily, making our hybrid model a useful building block for downstream applications of probabilistic deep 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。

↑