Improving VAE generations of multimodal data through data-dependent conditional priors

Lavda, Frantzeska; Gregorova, Magda; Kalousis, Alexandros

doi:10.48550/arxiv.1911.10885

Cited by 2 publications

(2 citation statements)

References 0 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Seeking for an optimal exploitation of the entire code space C, fixed priors in the form of diffuse uniforms are chosen here in agreement with the relevant literature [43]. Following the notation in [37], these represent the maximum entropy configuration for the discrete case…”

Section: A Probabilistic Discretization Of the Latentsmentioning

confidence: 99%

GenRadar: Self-supervised Probabilistic Camera Synthesis based on Radar Frequencies

Ditzel¹,

Dietmayer²

2021

Preprint

View full text Add to dashboard Cite

Autonomous systems require a continuous and dependable environment perception for navigation and decisionmaking, which is best achieved by combining different sensor types. Radar continues to function robustly in compromised circumstances in which cameras become impaired, guaranteeing a steady inflow of information. Yet, camera images provide a more intuitive and readily applicable impression of the world. This work combines the complementary strengths of both sensor types in a unique self-learning fusion approach for a probabilistic scene reconstruction in adverse surrounding conditions. After reducing the memory requirements of both high-dimensional measurements through a decoupled stochastic self-supervised compression technique, the proposed algorithm exploits similarities and establishes correspondences between both domains at different feature levels during training. Then, at inference time, relying exclusively on radio frequencies, the model successively predicts camera constituents in an autoregressive and selfcontained process. These discrete tokens are finally transformed back into an instructive view of the respective surrounding, allowing to visually perceive potential dangers for important tasks downstream.

show abstract

Section: A Probabilistic Discretization Of the Latentsmentioning

confidence: 99%

GenRadar: Self-supervised Probabilistic Camera Synthesis based on Radar Frequencies

Ditzel¹,

Dietmayer²

2021

Preprint

View full text Add to dashboard Cite

show abstract

“…Some models have considered them to be independent (Dupont (2018); Kim et al (2020)); hence, they cannot extract variations exclusive to particular classes. Some other models have considered all continuous variations to be class-dependent (Jiang et al (2017); Lavda et al (2019); Gao et al (2019)) and so unable to extract variations shared between different classes of data efficiently. Note that these models can learn shared factors of variation separately for each class using a part of the dataset, which will also waste the model's computation power.…”

Section: Related Workmentioning

confidence: 99%

Semi-Supervised Disentanglement of Class-Related and Class-Independent Factors in VAE

Sina¹,

Lotfi²,

Baghshah³

2021

Preprint

View full text Add to dashboard Cite

In recent years, extending variational autoencoder's framework to learn disentangled representations has received much attention. We address this problem by proposing a framework capable of disentangling class-related and class-independent factors of variation in data. Our framework employs an attention mechanism in its latent space in order to improve the process of extracting class-related factors from data. We also deal with the multimodality of data distribution by utilizing mixture models as learnable prior distributions, as well as incorporating the Bhattacharyya coefficient in the objective function to prevent highly overlapping mixtures. Our model's encoder is further trained in a semi-supervised manner, with a small fraction of labeled data, to improve representations' interpretability. Experiments show that our framework disentangles class-related and class-independent factors of variation and learns interpretable features. Moreover, we demonstrate our model's performance with quantitative and qualitative results on various datasets.

show abstract

Improving VAE generations of multimodal data through data-dependent conditional priors

Cited by 2 publications

References 0 publications

GenRadar: Self-supervised Probabilistic Camera Synthesis based on Radar Frequencies

GenRadar: Self-supervised Probabilistic Camera Synthesis based on Radar Frequencies

Semi-Supervised Disentanglement of Class-Related and Class-Independent Factors in VAE

Contact Info

Product

Resources

About