Résumé
To improve the ability of variational auto-encoders (VAE) to disentangle in the latent space, existing works mostly focus on enforcing the independence among the learned latent factors. However, the ability of these models to disentangle often decreases as the complexity of the generative factors increases. In this paper, we investigate the little-explored effect of the modeling capacity of a posterior density on the disentangling ability of the VAE. We note that the independence within and the complexity of the latent density are two different properties we constrain when regularizing the posterior density: while the former promotes the disentangling ability of VAE, the latter - if overly limited - creates an unnecessary competition with the data reconstruction objective in VAE. Therefore, if we preserve the independence but allow richer modeling capacity in the posterior density, we will lift this competition and thereby allow improved independence and data reconstruction at the same time. We investigate this theoretical intuition with a VAE that utilizes a non-parametric latent factor model, the Indian Buffet Process (IBP), as a latent density that is able to grow with the complexity of the data. Across two widely-used benchmark data sets (MNIST and dSprites) and two clinical data sets little explored for disentangled learning, we qualitatively and quantitatively demonstrated the improved disentangling performance of IBP-VAE over the state of the art. In the latter two clinical data sets riddled with complex factors of variations, we further demonstrated that unsupervised disentangling of nuisance factors via IBP-VAE - when combined with a supervised objective - can not only improve task accuracy in comparison to relevant supervised deep architectures, but also facilitate knowledge discovery related to task decision-making.
Langue d'origine | English |
---|---|
Titre de la publication principale | Proceedings - 19th IEEE International Conference on Data Mining, ICDM 2019 |
Éditeurs | Jianyong Wang, Kyuseok Shim, Xindong Wu |
Maison d'édition | Institute of Electrical and Electronics Engineers Inc. |
Pages | 1078-1083 |
Nombre de pages | 6 |
ISBN (électronique) | 9781728146034 |
DOI | |
Statut de publication | Published - nov. 2019 |
Événement | 19th IEEE International Conference on Data Mining, ICDM 2019 - Beijing, China Durée: nov. 8 2019 → nov. 11 2019 |
Séries de publication
Prénom | Proceedings - IEEE International Conference on Data Mining, ICDM |
---|---|
Volume | 2019-November |
ISSN (imprimé) | 1550-4786 |
Conference
Conference | 19th IEEE International Conference on Data Mining, ICDM 2019 |
---|---|
Pays/Territoire | China |
Ville | Beijing |
Période | 11/8/19 → 11/11/19 |
Note bibliographique
Publisher Copyright:© 2019 IEEE.
ASJC Scopus Subject Areas
- General Engineering