Unsupervised and Semi-supervised Learning with Categorical Generative Adversarial Networks

Abstract

In this paper we present a method for learning a discriminative classifier from unlabeled or partially labeled data. Our approach is based on an objective function that trades-off mutual information between observed examples and their predicted categorical class distribution, against robustness of the classifier to an adversarial generative model. The resulting algorithm can either be interpreted as a natural generalization of the generative adversarial networks (GAN) framework or as an extension of the regularized information maximization (RIM) framework to robust classification against an optimal adversary. We empirically evaluate our method - which we dub categorical generative adversarial networks (or CatGAN) - on synthetic data as well as on challenging image classification tasks, demonstrating the robustness of the learned classifiers. We further qualitatively assess the fidelity of samples generated by the adversarial generator that is learned alongside the discriminative classifier, and identify links between the CatGAN objective and discriminative clustering algorithms (such as RIM).

References in corpus (7)

• Improving neural networks by preventing co-adaptation of feature detectors
• Deep Generative Image Models using a Laplacian Pyramid of Adversarial Networks
• Theano: new features and speed improvements
• A note on the evaluation of generative models
• Generative Moment Matching Networks
• Learning with Pseudo-Ensembles
• No More Pesky Learning Rates

Cited by in corpus (127)

• Generative Adversarial Network in Medical Imaging: A Review
• NIPS 2016 Tutorial: Generative Adversarial Networks
• Adversarially Learned Inference
• Deep Unsupervised Clustering with Gaussian Mixture Variational Autoencoders
• How Generative Adversarial Networks and Their Variants Work: An Overview
• Lung and Pancreatic Tumor Characterization in the Deep Learning Era: Novel Supervised and Unsupervised Learning Approaches
• Learning Discrete Representations via Information Maximizing Self-Augmented Training
• A Review on Generative Adversarial Networks: Algorithms, Theory, and Applications
• An Overview of Deep Semi-Supervised Learning
• Good Semi-supervised Learning that Requires a Bad GAN
• Triple Generative Adversarial Nets
• Data Augmentation Using GANs
• Loss-Sensitive Generative Adversarial Networks on Lipschitz Densities
• BIVA: A Very Deep Hierarchy of Latent Variables for Generative Modeling
• Unsupervised Learning for Cell-level Visual Representation in Histopathology Images with Generative Adversarial Networks
• Semi-supervised Learning with GANs: Manifold Invariance with Improved Inference
• Deep Divergence-Based Approach to Clustering
• Feature-map-level Online Adversarial Knowledge Distillation
• Triangle Generative Adversarial Networks
• Subsurface structure analysis using computational interpretation and learning: A visual signal processing perspective
• Generalizing semi-supervised generative adversarial networks to regression using feature contrasting
• FrankenGAN: Guided Detail Synthesis for Building Mass-Models Using Style-Synchonized GANs
• Towards Utilizing Unlabeled Data in Federated Learning: A Survey and Prospective
• Semi and Weakly Supervised Semantic Segmentation Using Generative Adversarial Network
• Improving the Improved Training of Wasserstein GANs: A Consistency Term and Its Dual Effect
• Adversarial Dropout Regularization
• ExtremeWeather: A large-scale climate dataset for semi-supervised detection, localization, and understanding of extreme weather events
• AutoEmbedder: A semi-supervised DNN embedding system for clustering
• Activation Maximization Generative Adversarial Nets
• Conservative-Progressive Collaborative Learning for Semi-supervised Semantic Segmentation
• Augmented Cyclic Adversarial Learning for Low Resource Domain Adaptation
• Deep Discriminative Clustering Analysis
• Semi-supervised Conditional GANs
• Gaussian Mixture Generative Adversarial Networks for Diverse Datasets, and the Unsupervised Clustering of Images
• Novelty Detection with GAN
• Conditional Adversarial Network for Semantic Segmentation of Brain Tumor
• FSDR: Frequency Space Domain Randomization for Domain Generalization
• Structured Generative Adversarial Networks
• Boosting Deep Learning Risk Prediction with Generative Adversarial Networks for Electronic Health Records
• Smooth Neighbors on Teacher Graphs for Semi-supervised Learning
• DiscrimNet: Semi-Supervised Action Recognition from Videos using Generative Adversarial Networks
• Unsupervised Semantic Aggregation and Deformable Template Matching for Semi-Supervised Learning
• Semi-Conditional Normalizing Flows for Semi-Supervised Learning
• Dual Student: Breaking the Limits of the Teacher in Semi-supervised Learning
• Semi-supervised Text Regression with Conditional Generative Adversarial Networks
• CariGAN: Caricature Generation through Weakly Paired Adversarial Learning
• A Review of Semi Supervised Learning Theories and Recent Advances
• ArtGAN: Artwork Synthesis with Conditional Categorical GANs
• Hierarchical Adversarially Learned Inference
• Minimax Active Learning
• Robust Learning Under Label Noise With Iterative Noise-Filtering
• When Semi-Supervised Learning Meets Transfer Learning: Training Strategies, Models and Datasets
• Not to Cry Wolf: Distantly Supervised Multitask Learning in Critical Care
• Disentangled Variational Auto-Encoder for Semi-supervised Learning
• Effect of The Latent Structure on Clustering with GANs
• Aspect-augmented Adversarial Networks for Domain Adaptation
• Semi-Supervised Generation with Cluster-aware Generative Models
• Attribute Guided Unpaired Image-to-Image Translation with Semi-supervised Learning
• Is Discriminator a Good Feature Extractor?
• Disentangling Style and Content in Anime Illustrations
• Triple Generative Adversarial Networks
• MiniMax Entropy Network: Learning Category-Invariant Features for Domain Adaptation
• Understanding the Effectiveness of Lipschitz-Continuity in Generative Adversarial Nets
• Self-Supervised Relative Depth Learning for Urban Scene Understanding
• Boosting Noise Robustness of Acoustic Model via Deep Adversarial Training
• Weakly Supervised Clustering by Exploiting Unique Class Count
• Competing Against Equilibria in Zero-Sum Games with Evolving Payoffs
• Memory-Efficient Semi-Supervised Continual Learning: The World is its Own Replay Buffer
• Improving ClusterGAN Using Self-Augmented Information Maximization of Disentangling Latent Spaces
• Improving Deep Image Clustering With Spatial Transformer Layers
• An Adversarial Perturbation Oriented Domain Adaptation Approach for Semantic Segmentation
• A-Fast-RCNN: Hard Positive Generation via Adversary for Object Detection
• Bayesian Conditional Generative Adverserial Networks
• Countering Noisy Labels By Learning From Auxiliary Clean Labels
• IAN: Combining Generative Adversarial Networks for Imaginative Face Generation
• Manifold Adversarial Learning
• Efficient Clustering for Stretched Mixtures: Landscape and Optimality
• Predicting Opioid Relapse Using Social Media Data
• Unsupervised Learning using Pretrained CNN and Associative Memory Bank
• Semi-Supervised Learning Enabled by Multiscale Deep Neural Network Inversion
• Improved ArtGAN for Conditional Synthesis of Natural Image and Artwork
• Learning Deep Representations Using Convolutional Auto-encoders with Symmetric Skip Connections
• Semi-Supervised Learning with the Deep Rendering Mixture Model
• Modeling EEG data distribution with a Wasserstein Generative Adversarial Network to predict RSVP Events
• Removing Class Imbalance using Polarity-GAN: An Uncertainty Sampling Approach
• Mode matching in GANs through latent space learning and inversion
• Joint Learning of Generative Translator and Classifier for Visually Similar Classes
• Gotta Adapt 'Em All: Joint Pixel and Feature-Level Domain Adaptation for Recognition in the Wild
• Sparsely Grouped Multi-task Generative Adversarial Networks for Facial Attribute Manipulation
• Semi-supervised Learning using Adversarial Training with Good and Bad Samples
• PC-GAIN: Pseudo-label Conditional Generative Adversarial Imputation Networks for Incomplete Data
• Label Prediction Framework for Semi-Supervised Cross-Modal Retrieval
• SaaS: Speed as a Supervisor for Semi-supervised Learning
• Domain-Adversarial Training of Self-Attention Based Networks for Land Cover Classification using Multi-temporal Sentinel-2 Satellite Imagery
• Ensemble Manifold Segmentation for Model Distillation and Semi-supervised Learning
• Semi-Supervised Learning via New Deep Network Inversion
• Mixture of Spectral Generative Adversarial Networks for Imbalanced Hyperspectral Image Classification
• Reliable Semi-Supervised Learning when Labels are Missing at Random
• Semi-supervised learning for medical image classification using imbalanced training data
• Disentangled Representation Learning with Information Maximizing Autoencoder
• Multi-level Feature Learning on Embedding Layer of Convolutional Autoencoders and Deep Inverse Feature Learning for Image Clustering
• Weakly-supervised Generative Adversarial Networks for medical image classification
• GMAIR: Unsupervised Object Detection Based on Spatial Attention and Gaussian Mixture
• Understanding Natural Language Instructions for Fetching Daily Objects Using GAN-Based Multimodal Target-Source Classification
• A Multimodal Classifier Generative Adversarial Network for Carry and Place Tasks from Ambiguous Language Instructions
• PI-GAN: Learning Pose Independent representations for multiple profile face synthesis
• Unsupervised Representation Adversarial Learning Network: from Reconstruction to Generation
• Semi-Supervised Learning with IPM-based GANs: an Empirical Study
• Weakly Supervised Object Localization and Detection: A Survey
• Interpretable Deep Neural Networks for Facial Expression and Dimensional Emotion Recognition in-the-wild
• DINO: A Conditional Energy-Based GAN for Domain Translation
• Semi-Supervised Self-Growing Generative Adversarial Networks for Image Recognition
• ErGAN: Generative Adversarial Networks for Entity Resolution
• High Diversity Attribute Guided Face Generation with GANs
• Neural Adversarial Training for Semi-supervised Japanese Predicate-argument Structure Analysis
• Semi-Supervised Deep Learning Using Improved Unsupervised Discriminant Projection
• Adversarial Code Learning for Image Generation
• Selective Sampling and Mixture Models in Generative Adversarial Networks
• Virtual Conditional Generative Adversarial Networks
• Domain Adaptive Semantic Segmentation without Source Data
• An empirical study of domain-agnostic semi-supervised learning via energy-based models: joint-training and pre-training
• Grounded Language Understanding for Manipulation Instructions Using GAN-Based Classification
• Deep Dive into Semi-Supervised ELBO for Improving Classification Performance
• Linear Discriminant Generative Adversarial Networks
• A Self-Training Method for Semi-Supervised GANs
• Recurrent Deep Divergence-based Clustering for simultaneous feature learning and clustering of variable length time series
• Towards Stable Adversarial Feature Learning for LiDAR based Loop Closure Detection