Semi-Supervised Learning with Normalizing Flows

Normalizing flows transform a latent distribution through an invertible neural network for a flexible and pleasingly simple approach to generative modelling, while preserving an exact likelihood. We propose FlowGMM, an end-to-end approach to generative semi supervised learning with normalizing flows, using a latent Gaussian mixture model. FlowGMM is distinct in its simplicity, unified treatment of labelled and unlabelled data with an exact likelihood, interpretability, and broad applicability beyond image data. We show promising results on a wide range of applications, including AG-News and Yahoo Answers text data, tabular data, and semi-supervised image classification. We also show that FlowGMM can discover interpretable structure, provide real-time optimization-free feature visualizations, and specify well calibrated predictive distributions.

PDF Abstract ICML 2020 PDF

Datasets


Task Dataset Model Metric Name Metric Value Global Rank Result Benchmark
Semi-Supervised Text Classification AG News (200 Labels) 3 Layer MLP Accuracy (%) 77.5 # 3
Semi-Supervised Text Classification AG News (200 Labels) Pi Model Accuracy (%) 80.2 # 2
Semi-Supervised Text Classification AG News (200 Labels) FlowGMM Accuracy (%) 82.1 # 1
Semi-Supervised Text Classification Yahoo! Answers (800 Labels) 3 Layer MLP Accuracy (%) 55.7 # 3
Semi-Supervised Text Classification Yahoo! Answers (800 Labels) Pi Model Accuracy (%) 56.3 # 2
Semi-Supervised Text Classification Yahoo! Answers (800 Labels) FlowGMM Accuracy (%) 57.9 # 1

Methods


No methods listed for this paper. Add relevant methods here