Learning Disentangled Representations in the Imaging Domain

Liu, Xiao; Sanchez, Pedro; Thermos, Spyridon; O'Neil, Alison Q.; Tsaftaris, Sotirios A.

doi:10.48550/arxiv.2108.12043

Cited by 2 publications

(2 citation statements)

References 116 publications

(183 reference statements)

Supporting

Mentioning

Contrasting

Order By: Relevance

“…A latent representation is disentangled if each dimension in the latent space is sensitive to one generative factor and comparably invariant to the changes in the other factors (Liu et al, 2021). Such a disentangled representation is a great asset for interpretability.…”

Section: Variational Autoencoder (Vae) and β-Vaementioning

confidence: 99%

Attri-VAE: attribute-based, disentangled and interpretable representations of medical images with variational autoencoders

Çetin¹,

Cámara²,

Ballester³

2022

Preprint

View full text Add to dashboard Cite

Deep learning (DL) methods where interpretability is intrinsically considered as part of the model are required to better understand the relationship of clinical and imaging-based attributes with DL outcomes, thus facilitating their use in reasoning medical decisions. Latent space representations built with variational autoencoders (VAE) do not ensure individual control of data attributes. Attribute-based methods enforcing attribute disentanglement have been proposed in the literature for classical computer vision tasks in benchmark data. In this paper, we propose a VAE approach, the Attri-VAE, that includes an attribute regularization term to associate clinical and medical imaging attributes with different regularized dimensions in the generated latent space, enabling a better disentangled interpretation of the attributes. Furthermore, the generated attention maps explained the attribute encoding in the regularized latent space dimensions. The Attri-VAE approach analyzed healthy and myocardial infarction patients with clinical, cardiac morphology, and radiomics attributes. The proposed model provided an excellent trade-off between reconstruction fidelity, disentanglement, and interpretability, outperforming state-of-the-art VAE approaches according to several quantitative metrics. The resulting latent space allowed the generation of realistic synthetic data in the trajectory between two distinct input samples or along a specific attribute dimension to better interpret changes between different cardiac conditions.

show abstract

Section: Variational Autoencoder (Vae) and β-Vaementioning

confidence: 99%

Attri-VAE: attribute-based, disentangled and interpretable representations of medical images with variational autoencoders

Çetin¹,

Cámara²,

Ballester³

2022

Preprint

View full text Add to dashboard Cite

show abstract

“…Learning independent and semantic representations whose individual dimensions have interpretable meaning is usually referred to as disentangled representations learning (DRL) [2], [9], [10]. On the other hand, disentangled representation learning [11] aims to learn the representation of the underlying explainable factors behind the observed data and it is considered one of the possible ways for AI to fundamentally understand the world.…”

Section: Introductionmentioning

confidence: 99%

Towards Explainability of the Latent Space by Disentangled Representation Learning

Namatēvs,

Ņikuļins,

Slaidiņa

et al. 2023

ITMS

View full text Add to dashboard Cite

Deep neural networks are widely used in computer vision for image classification, segmentation and generation. They are also often criticised as “black boxes” because their decision-making process is often not interpretable by humans. However, learning explainable representations that explicitly disentangle the underlying mechanisms that structure observational data is still a challenge. To further explore the latent space and achieve generic processing, we propose a pipeline for discovering the explainable directions in the latent space of generative models. Since the latent space contains semantically meaningful directions and can be explained, we propose a pipeline to fully resolve the representation of the latent space. It consists of a Dirichlet encoder, conditional deterministic diffusion, a group-swap and a latent traversal module. We believe that this study provides an insight into the advancement of research explaining the disentanglement of neural networks in the community.

show abstract

Learning Disentangled Representations in the Imaging Domain

Cited by 2 publications

References 116 publications

Attri-VAE: attribute-based, disentangled and interpretable representations of medical images with variational autoencoders

Attri-VAE: attribute-based, disentangled and interpretable representations of medical images with variational autoencoders

Towards Explainability of the Latent Space by Disentangled Representation Learning

Contact Info

Product

Resources

About