Gradient Boosted Normalizing Flows

Abstract

By chaining a sequence of differentiable invertible transformations, normalizing flows (NF) provide an expressive method of posterior approximation, exact density evaluation, and sampling. The trend in normalizing flow literature has been to devise deeper, more complex transformations to achieve greater flexibility. We propose an alternative: Gradient Boosted Normalizing Flows (GBNF) model a density by successively adding new NF components with gradient boosting. Under the boosting framework, each new NF component optimizes a sample weighted likelihood objective, resulting in new components that are fit to the residuals of the previously trained components. The GBNF formulation results in a mixture model structure, whose flexibility increases as more components are added. Moreover, GBNFs offer a wider, as opposed to strictly deeper, approach that improves existing NFs at the cost of additional training---not more complex transformations. We demonstrate the effectiveness of this technique for density estimation and, by coupling GBNF with a variational autoencoder, generative modeling of images. Our results show that GBNFs outperform their non-boosted analog, and, in some cases, produce better results with smaller, simpler flows.

arXiv commentAppearing in the 34th Conference on Neural Information Processing Systems (NeurIPS 2020), Vancouver, Canada

References in corpus (29)

• Auto-Encoding Variational Bayes
• NICE: Non-linear Independent Components Estimation
• Variational Inference with Normalizing Flows
• Density estimation using Real NVP
• Normalizing Flows for Probabilistic Modeling and Inference
• MADE: Masked Autoencoder for Distribution Estimation
• Variational Lossy Autoencoder
• Flow++: Improving Flow-Based Generative Models with Variational Dequantization and Architecture Design
• Residual Flows for Invertible Generative Modeling
• Cyclical Annealing Schedule: A Simple Approach to Mitigating KL Vanishing
• AdaGAN: Boosting Generative Models
• Sylvester Normalizing Flows for Variational Inference
• RNADE: The real-valued neural autoregressive density-estimator
• On Unifying Deep Generative Models
• Improving Variational Auto-Encoders using Householder Flow
• Variational Boosting: Iteratively Refining Posterior Approximations
• Boosting Variational Inference
• Sum-of-Squares Polynomial Flow
• Online Gradient Boosting
• Continuous-Time Flows for Efficient Inference and Density Estimation
• Block Neural Autoregressive Flow
• Relaxing Bijectivity Constraints with Continuously Indexed Normalising Flows
• Boosting Variational Inference: an Optimization Perspective
• VFlow: More Expressive Generative Flows with Variational Data Augmentation
• Improving Explorability in Variational Inference with Annealed Variational Objectives
• Universal Boosting Variational Inference
• Boosted Generative Models
• Deep Diffeomorphic Normalizing Flows
• Boosted Density Estimation Remastered