VFlow: More Expressive Generative Flows with Variational Data Augmentation

Abstract

Generative flows are promising tractable models for density modeling that define probabilistic distributions with invertible transformations. However, tractability imposes architectural constraints on generative flows, making them less expressive than other types of generative models. In this work, we study a previously overlooked constraint that all the intermediate representations must have the same dimensionality with the original data due to invertibility, limiting the width of the network. We tackle this constraint by augmenting the data with some extra dimensions and jointly learning a generative flow for augmented data as well as the distribution of augmented dimensions under a variational inference framework. Our approach, VFlow, is a generalization of generative flows and therefore always performs better. Combining with existing generative flows, VFlow achieves a new state-of-the-art 2.98 bits per dimension on the CIFAR-10 dataset and is more compact than previous models to reach similar modeling quality.

References in corpus (15)

• Auto-Encoding Variational Bayes
• GANs Trained by a Two Time-Scale Update Rule Converge to a Local Nash Equilibrium
• Wide Residual Networks
• Pixel Recurrent Neural Networks
• NICE: Non-linear Independent Components Estimation
• Variational Inference with Normalizing Flows
• Variational Lossy Autoencoder
• Flow++: Improving Flow-Based Generative Models with Variational Dequantization and Architecture Design
• Residual Flows for Invertible Generative Modeling
• WAIC, but Why? Generative Ensembles for Robust Anomaly Detection
• PixelVAE: A Latent Variable Model for Natural Images
• Emerging Convolutions for Generative Normalizing Flows
• Hybrid Models with Deep and Invertible Features
• Augmented Normalizing Flows: Bridging the Gap Between Generative Flows and Latent Variable Models
• Variational Autoencoders with Normalizing Flow Decoders

Cited by in corpus (8)

• NVAE: A Deep Hierarchical Variational Autoencoder
• Deep Generative Modelling: A Comparative Review of VAEs, GANs, Normalizing Flows, Energy-Based and Autoregressive Models
• Maximum Likelihood Training of Score-Based Diffusion Models
• SurVAE Flows: Surjections to Bridge the Gap between VAEs and Flows
• Score-based Generative Modeling in Latent Space
• Likelihood Training of Schrödinger Bridge using Forward-Backward SDEs Theory
• Decoupling Global and Local Representations via Invertible Generative Flows
• Gradient Boosted Normalizing Flows