Neural Spline Flows
Neural Spline Flows
复制标题
DOI：
--
复制
发表时间：
2019-06
期刊：
影响因子：
--
通讯作者：
Conor Durkan;Artur Bekasov;Iain Murray;G. Papamakarios
中科院分区：
文献类型：
--
作者：
Conor Durkan;Artur Bekasov;Iain Murray;G. Papamakarios
归一化流将复杂的概率密度建模为简单基础密度的可逆变换。基于耦合或自回归变换的流都提供精确的密度评估和采样，但依赖于一个容易可逆的元素级变换的参数化，其选择决定了这些模型的灵活性。基于最近的工作，我们提出了一个基于单调有理二次样条的完全可微模块，它增强了耦合和自回归变换的灵活性，同时保留了分析可逆性。我们证明了神经样条流提高了密度估计，变分推理和图像的生成建模。
A normalizing flow models a complex probability density as an invertible transformation of a simple base density. Flows based on either coupling or autoregressive transforms both offer exact density evaluation and sampling, but rely on the parameterization of an easily invertible elementwise transformation, whose choice determines the flexibility of these models. Building upon recent work, we propose a fully-differentiable module based on monotonic rational-quadratic splines, which enhances the flexibility of both coupling and autoregressive transforms while retaining analytic invertibility. We demonstrate that neural spline flows improve density estimation, variational inference, and generative modeling of images.