RG-Flow: A hierarchical and explainable flow model based on renormalization group and sparse prior

Abstract

Flow-based generative models have become an important class of unsupervised learning approaches. In this work, we incorporate the key ideas of renormalization group (RG) and sparse prior distribution to design a hierarchical flow-based generative model, RG-Flow, which can separate information at different scales of images and extract disentangled representations at each scale. We demonstrate our method on synthetic multi-scale image datasets and the CelebA dataset, showing that the disentangled representations enable semantic manipulation and style mixing of the images at different scales. To visualize the latent representations, we introduce receptive fields for flow-based models and show that the receptive fields of RG-Flow are similar to those of convolutional neural networks. In addition, we replace the widely adopted isotropic Gaussian prior distribution by the sparse Laplacian distribution to further enhance the disentanglement of representations. From a theoretical perspective, our proposed method has $O(\log L)$ complexity for inpainting of an image with edge length $L$, compared to previous generative models with $O(L^2)$ complexity.

arXiv comment31 pages, 20 figures, 3 tables

References in corpus (38)

• Auto-Encoding Variational Bayes
• Decoupled Weight Decay Regularization
• Quantum Convolutional Neural Networks
• NICE: Non-linear Independent Components Estimation
• A class of quantum many-body states that can be efficiently simulated
• Entanglement Renormalization and Holography
• Holographic quantum error-correcting codes: Toy models for the bulk/boundary correspondence
• Challenging Common Assumptions in the Unsupervised Learning of Disentangled Representations
• Disentangling by Factorising
• Texture Networks: Feed-forward Synthesis of Textures and Stylized Images
• Why does deep and cheap learning work so well?
• NVAE: A Deep Hierarchical Variational Autoencoder
• Mutual Information, Neural Networks and the Renormalization Group
• Residual Flows for Invertible Generative Modeling
• An exact mapping between the Variational Renormalization Group and Deep Learning
• Neural Network Renormalization Group
• Discovering Hidden Factors of Variation in Deep Networks
• Texture Synthesis with Spatial Generative Adversarial Networks
• Exact holographic mapping and emergent space-time geometry
• A class of highly entangled many-body states that can be efficiently simulated
• Machine Learning Spatial Geometry from Entanglement Features
• Normalizing Flows on Tori and Spheres
• Deep Learning and Holographic QCD
• Entanglement Holographic Mapping of Many-Body Localized System by Spectrum Bifurcation Renormalization Group
• AdS/CFT as a deep Boltzmann machine
• Deep Learning and AdS/QCD
• Machine Learning Holographic Mapping by Neural Network Renormalization Group
• Holography as deep learning
• Emerging Convolutions for Generative Normalizing Flows
• Exact holographic mapping in free fermion systems
• Flows for simultaneous manifold learning and density estimation
• The renormalisation group via statistical inference
• Understanding Anomaly Detection with Deep Invertible Networks through Hierarchies of Distributions and Features
• Optimal Renormalization Group Transformation from Information Theory
• A Spectral Regularizer for Unsupervised Disentanglement
• Word Embedding Visualization Via Dictionary Learning
• Transflow Learning: Repurposing Flow Models Without Retraining
• Criticality & Deep Learning II: Momentum Renormalisation Group

Cited by in corpus (3)

• Neural Information Squeezer for Causal Emergence
• Differentiable Programming of Isometric Tensor Networks
• Learning Non-linear Wavelet Transformation via Normalizing Flow