Tractable Density Estimation on Learned Manifolds with Conformal Embedding Flows

Abstract

Normalizing flows are generative models that provide tractable density estimation via an invertible transformation from a simple base distribution to a complex target distribution. However, this technique cannot directly model data supported on an unknown low-dimensional manifold, a common occurrence in real-world domains such as image data. Recent attempts to remedy this limitation have introduced geometric complications that defeat a central benefit of normalizing flows: exact density estimation. We recover this benefit with Conformal Embedding Flows, a framework for designing flows that learn manifolds with tractable densities. We argue that composing a standard flow with a trainable conformal embedding is the most natural way to model manifold-supported data. To this end, we present a series of conformal building blocks and apply them in experiments with synthetic and real-world data to demonstrate that flows can model manifold-supported distributions without sacrificing tractable likelihoods.

arXiv commentNeurIPS 2021 Camera-Ready. Code: https://github.com/layer6ai-labs/CEF

References in corpus (25)

• GANs Trained by a Two Time-Scale Update Rule Converge to a Local Nash Equilibrium
• Large Scale GAN Training for High Fidelity Natural Image Synthesis
• Pixel Recurrent Neural Networks
• NICE: Non-linear Independent Components Estimation
• Variational Inference with Normalizing Flows
• Generative Modeling by Estimating Gradients of the Data Distribution
• Generalization and Equilibrium in Generative Adversarial Nets (GANs)
• Assessing Generative Models via Precision and Recall
• Image Representations Learned With Unsupervised Pre-Training Contain Human-like Biases
• Orthogonal Weight Normalization: Solution to Optimization over Multiple Dependent Stiefel Manifolds in Deep Neural Networks
• Cheap Orthogonal Constraints in Neural Networks: A Simple Parametrization of the Orthogonal and Unitary Group
• Improving Variational Auto-Encoders using Householder Flow
• Normalizing Flows on Tori and Spheres
• Flow-GAN: Combining Maximum Likelihood and Adversarial Learning in Generative Models
• Generalized BackPropagation, Étude De Cas: Orthogonality
• Optimization on Submanifolds of Convolution Kernels in CNNs
• Trivializations for Gradient-Based Optimization on Manifolds
• Riemannian Continuous Normalizing Flows
• Deep Isometric Learning for Visual Recognition
• LOCA: LOcal Conformal Autoencoder for standardized data coordinates
• Regularized Autoencoders via Relaxed Injective Probability Flow
• Normalizing Flows Across Dimensions
• Rectangular Flows for Manifold Learning
• Fair Normalizing Flows
• Generalization of the Change of Variables Formula with Applications to Residual Flows

Cited by in corpus (1)

• Rectangular Flows for Manifold Learning