Cubic-Spline Flows

Abstract

A normalizing flow models a complex probability density as an invertible transformation of a simple density. The invertibility means that we can evaluate densities and generate samples from a flow. In practice, autoregressive flow-based models are slow to invert, making either density estimation or sample generation slow. Flows based on coupling transforms are fast for both tasks, but have previously performed less well at density estimation than autoregressive flows. We stack a new coupling transform, based on monotonic cubic splines, with LU-decomposed linear layers. The resulting cubic-spline flow retains an exact one-pass inverse, can be used to generate high-quality images, and closes the gap with autoregressive flows on a suite of density-estimation tasks.

Appeared at the 1st Workshop on Invertible Neural Networks and Normalizing Flows at ICML 2019

Cited by in corpus (21)

• Normalizing Flows for Probabilistic Modeling and Inference
• Deep Generative Modelling: A Comparative Review of VAEs, GANs, Normalizing Flows, Energy-Based and Autoregressive Models
• Generative Networks for Precision Enthusiasts
• Efficient Modelling of Trivializing Maps for Lattice ${\varphi }^4$ Theory Using Normalizing Flows: A First Look at Scalability
• Continuous and Distribution-free Probabilistic Wind Power Forecasting: A Conditional Normalizing Flow Approach
• A neural simulation-based inference approach for characterizing the Galactic Center $\gamma$-ray excess
• Neural Spline Flows
• The Landscape of Unfolding with Machine Learning
• Normalizing Flows for High-Dimensional Detector Simulations
• TraDE: Transformers for Density Estimation
• Optimal, fast, and robust inference of reionization-era cosmology with the 21cmPIE-INN
• Invertible Generative Modeling using Linear Rational Splines
• Deep Conditional Transformation Models
• Identifying through Flows for Recovering Latent Representations
• Robust normalizing flows using Bernstein-type polynomials
• The Convolution Exponential and Generalized Sylvester Flows
• Likelihood Contribution based Multi-scale Architecture for Generative Flows
• BitHEP -- The Limits of Low-Precision ML in HEP
• Diffeomorphic Transformations for Time Series Analysis: An Efficient Approach to Nonlinear Warping
• Universal Joint Approximation of Manifolds and Densities by Simple Injective Flows
• Improving the expressiveness of neural vocoding with non-affine Normalizing Flows