Learning Latent Space Energy-Based Prior Model

Abstract

We propose to learn energy-based model (EBM) in the latent space of a generator model, so that the EBM serves as a prior model that stands on the top-down network of the generator model. Both the latent space EBM and the top-down network can be learned jointly by maximum likelihood, which involves short-run MCMC sampling from both the prior and posterior distributions of the latent vector. Due to the low dimensionality of the latent space and the expressiveness of the top-down network, a simple EBM in latent space can capture regularities in the data effectively, and MCMC sampling in latent space is efficient and mixes well. We show that the learned model exhibits strong performances in terms of image and text generation and anomaly detection. The one-page code can be found in supplementary materials.

NeurIPS 2020 Camera-Ready

References in corpus (23)

• Auto-Encoding Variational Bayes
• Large Scale GAN Training for High Fidelity Natural Image Synthesis
• Neural Discrete Representation Learning
• Variational Inference with Normalizing Flows
• Efficient GAN-Based Anomaly Detection
• A Connection between Generative Adversarial Networks, Inverse Reinforcement Learning, and Energy-Based Models
• Better Mixing via Deep Representations
• Optimizing the Latent Space of Generative Networks
• Wasserstein Auto-Encoders
• A Theory of Generative ConvNet
• Implicit Generation and Generalization in Energy-Based Models
• Learning Non-Convergent Non-Persistent Short-Run MCMC Toward Energy-Based Model
• Maximum Entropy Generators for Energy-Based Models
• Alternating Back-Propagation for Generator Network
• Discrete Variational Autoencoders
• DVAE++: Discrete Variational Autoencoders with Overlapping Transformations
• Introspective Classification with Convolutional Nets
• Eval all, trust a few, do wrong to none: Comparing sentence generation models
• DVAE#: Discrete Variational Autoencoders with Relaxed Boltzmann Priors
• Learning FRAME Models Using CNN Filters
• Calibrating Energy-based Generative Adversarial Networks
• Learning Latent Space Energy-Based Prior Model for Molecule Generation
• Semi-supervised Learning by Latent Space Energy-Based Model of Symbol-Vector Coupling

Cited by in corpus (4)

• A Contrastive Learning Approach for Training Variational Autoencoder Priors
• Learning Latent Space Energy-Based Prior Model for Molecule Generation
• A prior-based approximate latent Riemannian metric
• Semi-supervised Learning by Latent Space Energy-Based Model of Symbol-Vector Coupling