AdaGAN: Boosting Generative Models

Abstract

Generative Adversarial Networks (GAN) (Goodfellow et al., 2014) are an effective method for training generative models of complex data such as natural images. However, they are notoriously hard to train and can suffer from the problem of missing modes where the model is not able to produce examples in certain regions of the space. We propose an iterative procedure, called AdaGAN, where at every step we add a new component into a mixture model by running a GAN algorithm on a reweighted sample. This is inspired by boosting algorithms, where many potentially weak individual predictors are greedily aggregated to form a strong composite predictor. We prove that such an incremental procedure leads to convergence to the true distribution in a finite number of steps if each step is optimal, and convergence at an exponential rate otherwise. We also illustrate experimentally that this procedure addresses the problem of missing modes.

Updated with MNIST pictures and discussions + Unrolled GAN experiments

Cited by in corpus (48)

• Generative Adversarial Networks: An Overview
• A Review on Generative Adversarial Networks: Algorithms, Theory, and Applications
• MD-GAN: Multi-Discriminator Generative Adversarial Networks for Distributed Datasets
• Ten Years of Generative Adversarial Nets (GANs): A survey of the state-of-the-art
• Stabilizing Generative Adversarial Networks: A Survey
• Spatial Evolutionary Generative Adversarial Networks
• Style Aggregated Network for Facial Landmark Detection
• BourGAN: Generative Networks with Metric Embeddings
• Adversarial Training in Affective Computing and Sentiment Analysis: Recent Advances and Perspectives
• AdaGCN: Adaboosting Graph Convolutional Networks into Deep Models
• Deep Generative Models for Detector Signature Simulation: A Taxonomic Review
• Generative Adversarial Networks for Financial Trading Strategies Fine-Tuning and Combination
• Copulas as High-Dimensional Generative Models: Vine Copula Autoencoders
• A hybrid quantum-classical conditional generative adversarial network algorithm for human-centered paradigm in cloud
• Imperfect ImaGANation: Implications of GANs Exacerbating Biases on Facial Data Augmentation and Snapchat Selfie Lenses
• Generative Adversarial Networks (GANs): What it can generate and What it cannot?
• Group Ensemble: Learning an Ensemble of ConvNets in a single ConvNet
• Private Post-GAN Boosting
• Composite Functional Gradient Learning of Generative Adversarial Models
• MotionVideoGAN: A Novel Video Generator Based on the Motion Space Learned from Image Pairs
• Learning disconnected manifolds: a no GANs land
• Key Protected Classification for Collaborative Learning
• MCL-GAN: Generative Adversarial Networks with Multiple Specialized Discriminators
• Constrained Generative Adversarial Network Ensembles for Sharable Synthetic Data Generation
• Poincaré Recurrence, Cycles and Spurious Equilibria in Gradient-Descent-Ascent for Non-Convex Non-Concave Zero-Sum Games
• CDE-GAN: Cooperative Dual Evolution Based Generative Adversarial Network
• On Characterizing GAN Convergence Through Proximal Duality Gap
• Snore-GANs: Improving Automatic Snore Sound Classification with Synthesized Data
• Influence Estimation for Generative Adversarial Networks
• Deep Generative Learning via Variational Gradient Flow
• Latent Dirichlet Allocation in Generative Adversarial Networks
• A Deep Factorization of Style and Structure in Fonts
• GANs with Variational Entropy Regularizers: Applications in Mitigating the Mode-Collapse Issue
• Ensembles of GANs for synthetic training data generation
• Deconstructing Generative Adversarial Networks
• Robust GANs against Dishonest Adversaries
• Semi-Implicit Generative Model
• A Novel Measure to Evaluate Generative Adversarial Networks Based on Direct Analysis of Generated Images
• PAC-GAN: An Effective Pose Augmentation Scheme for Unsupervised Cross-View Person Re-identification
• Adversarial sampling of unknown and high-dimensional conditional distributions
• Student's t-Generative Adversarial Networks
• Integral Privacy for Sampling
• Boosting Generative Models by Leveraging Cascaded Meta-Models
• Virtual Conditional Generative Adversarial Networks
• Neural Network based Explicit Mixture Models and Expectation-maximization based Learning
• Greedy Convex Ensemble
• Quantifying point cloud realism through adversarially learned latent representations
• Gradient Boosted Normalizing Flows