Hierarchical Adversarially Learned Inference

TL;DR

This paper introduces a hierarchical generative model trained via adversarial learning that learns abstract, semantically meaningful representations, outperforming handcrafted features and recent supervised methods on CelebA and MNIST datasets.

Contribution

The paper presents a novel hierarchical adversarially learned inference model with a simple Markovian structure, enabling unsupervised learning of meaningful features and state-of-the-art semi-supervised classification.

Findings

Hierarchical structure supports learning of abstract representations.

Model outperforms handcrafted features on CelebA.

Achieves state-of-the-art semi-supervised MNIST classification.

Abstract

We propose a novel hierarchical generative model with a simple Markovian structure and a corresponding inference model. Both the generative and inference model are trained using the adversarial learning paradigm. We demonstrate that the hierarchical structure supports the learning of progressively more abstract representations as well as providing semantically meaningful reconstructions with different levels of fidelity. Furthermore, we show that minimizing the Jensen-Shanon divergence between the generative and inference network is enough to minimize the reconstruction error. The resulting semantically meaningful hierarchical latent structure discovery is exemplified on the CelebA dataset. There, we show that the features learned by our model in an unsupervised way outperform the best handcrafted features. Furthermore, the extracted features remain competitive when compared to several recent deep supervised approaches on an attribute prediction task on CelebA. Finally, we leverage the model's inference network to achieve state-of-the-art performance on a semi-supervised variant of the MNIST digit classification task.