Projects per year
Abstract
Learning disentangled representations requires either supervision or the introduction of specific model designs and learning constraints as biases. InfoGAN is a popular disentanglement framework that learns unsupervised disentangled representations by maximising the mutual information between latent representations and their corresponding generated images. Maximisation of mutual information is achieved by introducing an auxiliary network and training with a latent regression loss. In this short exploratory paper, we study the use of the Hilbert-Schmidt Independence Criterion (HSIC) to approximate mutual information between latent representation and image, termed HSIC-InfoGAN. Directly optimising the HSIC loss avoids the need for an additional auxiliary network. We qualitatively compare the level of disentanglement in each model, suggest a strategy to tune the hyperparameters of HSIC-InfoGAN, and discuss the potential of HSIC-InfoGAN for medical applications.
Original language | English |
---|---|
Title of host publication | Medical Applications with Disentanglements - First MICCAI Workshop, MAD 2022, Held in Conjunction with MICCAI 2022, Proceedings |
Editors | Jana Fragemann, Jianning Li, Jan Egger, Xiao Liu, Sotirios A. Tsaftaris, Jens Kleesiek |
Publisher | Springer |
Pages | 15-21 |
Number of pages | 7 |
ISBN (Print) | 9783031250453 |
DOIs | |
Publication status | Published - 1 Feb 2023 |
Event | 1st MICCAI Workshop on Medical Applications with Disentanglements, MAD 2022, held in conjunction with MICCAI 2022 - Singapore, Singapore Duration: 22 Sept 2022 → 22 Sept 2022 |
Publication series
Name | Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) |
---|---|
Volume | 13823 LNCS |
ISSN (Print) | 0302-9743 |
ISSN (Electronic) | 1611-3349 |
Conference
Conference | 1st MICCAI Workshop on Medical Applications with Disentanglements, MAD 2022, held in conjunction with MICCAI 2022 |
---|---|
Country/Territory | Singapore |
City | Singapore |
Period | 22/09/22 → 22/09/22 |
Keywords / Materials (for Non-textual outputs)
- Disentangled representation learning
- HSIC
- InfoGAN
Fingerprint
Dive into the research topics of 'HSIC-InfoGAN: Learning Unsupervised Disentangled Representations by Maximising Approximated Mutual Information'. Together they form a unique fingerprint.Projects
- 2 Finished