Adversarial Autoencoders

Abstract

In this paper, we propose the "adversarial autoencoder" (AAE), which is a probabilistic autoencoder that uses the recently proposed generative adversarial networks (GAN) to perform variational inference by matching the aggregated posterior of the hidden code vector of the autoencoder with an arbitrary prior distribution. Matching the aggregated posterior to the prior ensures that generating from any part of prior space results in meaningful samples. As a result, the decoder of the adversarial autoencoder learns a deep generative model that maps the imposed prior to the data distribution. We show how the adversarial autoencoder can be used in applications such as semi-supervised classification, disentangling style and content of images, unsupervised clustering, dimensionality reduction and data visualization. We performed experiments on MNIST, Street View House Numbers and Toronto Face datasets and show that adversarial autoencoders achieve competitive results in generative modeling and semi-supervised classification tasks.

References in corpus (3)

• Semi-Supervised Learning with Deep Generative Models
• Generative Moment Matching Networks
• Better Mixing via Deep Representations

Cited by in corpus (114)

• Unpaired Image-to-Image Translation using Cycle-Consistent Adversarial Networks
• Adversarial Uncertainty Quantification in Physics-Informed Neural Networks
• Augmented CycleGAN: Learning Many-to-Many Mappings from Unpaired Data
• Wasserstein Auto-Encoders
• DRIT++: Diverse Image-to-Image Translation via Disentangled Representations
• Getting High: High Fidelity Simulation of High Granularity Calorimeters with High Speed
• Variational Discriminator Bottleneck: Improving Imitation Learning, Inverse RL, and GANs by Constraining Information Flow
• Learning on Attribute-Missing Graphs
• Deep Learning and Knowledge-Based Methods for Computer Aided Molecular Design -- Toward a Unified Approach: State-of-the-Art and Future Directions
• Super-Resolution via Deep Learning
• From Variational to Deterministic Autoencoders
• Roundtrip: A Deep Generative Neural Density Estimator
• Age Progression/Regression by Conditional Adversarial Autoencoder
• From optimal transport to generative modeling: the VEGAN cookbook
• Deep Variational Canonical Correlation Analysis
• Information Constraints on Auto-Encoding Variational Bayes
• Variational Deep Embedding: An Unsupervised and Generative Approach to Clustering
• ChatPainter: Improving Text to Image Generation using Dialogue
• An analysis on the use of autoencoders for representation learning: fundamentals, learning task case studies, explainability and challenges
• Variational Inference using Implicit Distributions
• Counterfactual State Explanations for Reinforcement Learning Agents via Generative Deep Learning
• Adversarial Network Embedding
• Artificial Intelligence in Glioma Imaging: Challenges and Advances
• Inspect, Understand, Overcome: A Survey of Practical Methods for AI Safety
• An advanced hybrid deep adversarial autoencoder for parameterized nonlinear fluid flow modelling
• Stacked Generative Adversarial Networks
• Graph Normalizing Flows
• Adversarial Symmetric Variational Autoencoder
• Does Unsupervised Architecture Representation Learning Help Neural Architecture Search?
• Geometric Understanding of Deep Learning
• Eval all, trust a few, do wrong to none: Comparing sentence generation models
• A Tutorial on Deep Latent Variable Models of Natural Language
• Unsupervised Learning of Anomaly Detection from Contaminated Image Data using Simultaneous Encoder Training
• Adversarial Autoencoders with Constant-Curvature Latent Manifolds
• Conditional Adversarial Synthesis of 3D Facial Action Units
• Generative Modeling with Conditional Autoencoders: Building an Integrated Cell
• Latent Constraints: Learning to Generate Conditionally from Unconditional Generative Models
• Relevance Factor VAE: Learning and Identifying Disentangled Factors
• Encoding large scale cosmological structure with Generative Adversarial Networks
• The Deep Kernelized Autoencoder
• Multi-Modal Adversarial Autoencoders for Recommendations of Citations and Subject Labels
• Zero-Shot Visual Recognition using Semantics-Preserving Adversarial Embedding Networks
• KATE: K-Competitive Autoencoder for Text
• Modality to Modality Translation: An Adversarial Representation Learning and Graph Fusion Network for Multimodal Fusion
• Detection of Accounting Anomalies in the Latent Space using Adversarial Autoencoder Neural Networks
• Acoustic Leak Detection in Water Networks
• Ranking CGANs: Subjective Control over Semantic Image Attributes
• Perceptual Generative Autoencoders
• Task Specific Adversarial Cost Function
• Cross-language Learning with Adversarial Neural Networks: Application to Community Question Answering
• Invariant Representations from Adversarially Censored Autoencoders
• ODIN: Automated Drift Detection and Recovery in Video Analytics
• Bounded Information Rate Variational Autoencoders
• Primal-Dual Wasserstein GAN
• Variational Inference: A Unified Framework of Generative Models and Some Revelations
• Spatial Frequency Loss for Learning Convolutional Autoencoders
• An Improved Self-supervised GAN via Adversarial Training
• Disentangled Representation Learning with Wasserstein Total Correlation
• Disentangling and Learning Robust Representations with Natural Clustering
• Learning latent representations across multiple data domains using Lifelong VAEGAN
• Gaussian Copula Variational Autoencoders for Mixed Data
• MMGAN: Generative Adversarial Networks for Multi-Modal Distributions
• Persuasive Faces: Generating Faces in Advertisements
• Gaussian mixture models with Wasserstein distance
• Deep Cross-Modal Audio-Visual Generation
• Generative Model with Coordinate Metric Learning for Object Recognition Based on 3D Models
• Learning Compositional Visual Concepts with Mutual Consistency
• Human-interpretable model explainability on high-dimensional data
• Semi-Supervised Haptic Material Recognition for Robots using Generative Adversarial Networks
• Discriminator Feature-based Inference by Recycling the Discriminator of GANs
• Improving VAEs' Robustness to Adversarial Attack
• Cycle-Consistent Adversarial Learning as Approximate Bayesian Inference
• Learning a low dimensional manifold of real cancer tissue with PathologyGAN
• Importance Weighted Adversarial Variational Autoencoders for Spike Inference from Calcium Imaging Data
• Deep Generative Learning via Variational Gradient Flow
• Deterministic Decoding for Discrete Data in Variational Autoencoders
• Learning Implicit Generative Models with Theoretical Guarantees
• Plug and Play Autoencoders for Conditional Text Generation
• Hyperprior Induced Unsupervised Disentanglement of Latent Representations
• Class-Distinct and Class-Mutual Image Generation with GANs
• Deep Co-Attention Network for Multi-View Subspace Learning
• Analytical Moment Regularizer for Gaussian Robust Networks
• Semi-supervised Disentanglement with Independent Vector Variational Autoencoders
• 3D Shape Reconstruction from a Single 2D Image via 2D-3D Self-Consistency
• Novelty Detection via Robust Variational Autoencoding
• Anomaly Detection by Recombining Gated Unsupervised Experts
• A lower bound for the ELBO of the Bernoulli Variational Autoencoder
• Bayes-Factor-VAE: Hierarchical Bayesian Deep Auto-Encoder Models for Factor Disentanglement
• Generative Learning With Euler Particle Transport
• Attribute-Guided Sketch Generation
• Adversarially Learned Mixture Model
• Conformation Clustering of Long MD Protein Dynamics with an Adversarial Autoencoder
• Learning Controllable Disentangled Representations with Decorrelation Regularization
• Sliced Wasserstein Generative Models
• SimpleChrome: Encoding of Combinatorial Effects for Predicting Gene Expression
• Scalable and Effective Deep CCA via Soft Decorrelation
• TzK Flow - Conditional Generative Model
• DLGAN: Disentangling Label-Specific Fine-Grained Features for Image Manipulation
• Non-Asymptotic Error Bounds for Bidirectional GANs
• Semi-Supervised Disentanglement of Class-Related and Class-Independent Factors in VAE
• Blocked and Hierarchical Disentangled Representation From Information Theory Perspective
• MaskAAE: Latent space optimization for Adversarial Auto-Encoders
• A New Distribution on the Simplex with Auto-Encoding Applications
• Node Attribute Generation on Graphs
• Inducing Functions through Reinforcement Learning without Task Specification
• Supervised Topological Maps
• On-Demand Video Dispatch Networks: A Scalable End-to-End Learning Approach
• TzK: Flow-Based Conditional Generative Model
• WiSE-ALE: Wide Sample Estimator for Approximate Latent Embedding
• A Multi-Task Approach for Disentangling Syntax and Semantics in Sentence Representations
• Generative Adversarial Stacked Autoencoders
• NeurInt : Learning to Interpolate through Neural ODEs
• Toward Learning a Unified Many-to-Many Mapping for Diverse Image Translation
• Stacked Wasserstein Autoencoder