Generative-Discriminative Complementary Learning

Abstract

Majority of state-of-the-art deep learning methods are discriminative approaches, which model the conditional distribution of labels given inputs features. The success of such approaches heavily depends on high-quality labeled instances, which are not easy to obtain, especially as the number of candidate classes increases. In this paper, we study the complementary learning problem. Unlike ordinary labels, complementary labels are easy to obtain because an annotator only needs to provide a yes/no answer to a randomly chosen candidate class for each instance. We propose a generative-discriminative complementary learning method that estimates the ordinary labels by modeling both the conditional (discriminative) and instance (generative) distributions. Our method, we call Complementary Conditional GAN (CCGAN), improves the accuracy of predicting ordinary labels and can generate high-quality instances in spite of weak supervision. In addition to the extensive empirical studies, we also theoretically show that our model can retrieve the true conditional distribution from the complementarily-labeled data.

References in corpus (22)

• Conditional Generative Adversarial Nets
• Rethinking Atrous Convolution for Semantic Image Segmentation
• Deep Residual Learning for Image Recognition
• Geometric deep learning: going beyond Euclidean data
• Large Scale GAN Training for High Fidelity Natural Image Synthesis
• Spectral Normalization for Generative Adversarial Networks
• Self-Attention Generative Adversarial Networks
• Conditional Image Synthesis With Auxiliary Classifier GANs
• Semi-Supervised Learning with Deep Generative Models
• Improved Training of Wasserstein GANs
• Attention-based Deep Multiple Instance Learning
• Semi-Supervised Learning with Generative Adversarial Networks
• Training Convolutional Networks with Noisy Labels
• Multi-Instance Multi-Label Learning
• Good Semi-supervised Learning that Requires a Bad GAN
• Positive-Unlabeled Learning with Non-Negative Risk Estimator
• Semi-supervised Learning with GANs: Manifold Invariance with Improved Inference
• Learning From Noisy Singly-labeled Data
• Learning from Complementary Labels
• Robustness of Conditional GANs to Noisy Labels
• Twin Auxiliary Classifiers GAN
• Label-Noise Robust Generative Adversarial Networks

Cited by in corpus (2)

• Learning with Multiple Complementary Labels
• Strength from Weakness: Fast Learning Using Weak Supervision