Controlling Posterior Collapse by an Inverse Lipschitz Constraint on the Decoder Network
Fuente:
arXiv
Salvato in:
| Autori principali: | , , , , |
|---|---|
| Natura: | Preprint |
| Pubblicazione: |
2023
|
| Soggetti: | |
| Accesso online: | |
| Tags: |
Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
|
| _version_ | 1866913221657493504 |
|---|---|
| author | Kinoshita, Yuri Oono, Kenta Fukumizu, Kenji Yoshida, Yuichi Maeda, Shin-ichi |
| author_facet | Kinoshita, Yuri Oono, Kenta Fukumizu, Kenji Yoshida, Yuichi Maeda, Shin-ichi |
| contents | Variational autoencoders (VAEs) are one of the deep generative models that have experienced enormous success over the past decades. However, in practice, they suffer from a problem called posterior collapse, which occurs when the encoder coincides, or collapses, with the prior taking no information from the latent structure of the input data into consideration. In this work, we introduce an inverse Lipschitz neural network into the decoder and, based on this architecture, provide a new method that can control in a simple and clear manner the degree of posterior collapse for a wide range of VAE models equipped with a concrete theoretical guarantee. We also illustrate the effectiveness of our method through several numerical experiments. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2304_12770 |
| institution | arXiv |
| publishDate | 2023 |
| record_format | arxiv |
| spellingShingle | Controlling Posterior Collapse by an Inverse Lipschitz Constraint on the Decoder Network Kinoshita, Yuri Oono, Kenta Fukumizu, Kenji Yoshida, Yuichi Maeda, Shin-ichi Machine Learning Variational autoencoders (VAEs) are one of the deep generative models that have experienced enormous success over the past decades. However, in practice, they suffer from a problem called posterior collapse, which occurs when the encoder coincides, or collapses, with the prior taking no information from the latent structure of the input data into consideration. In this work, we introduce an inverse Lipschitz neural network into the decoder and, based on this architecture, provide a new method that can control in a simple and clear manner the degree of posterior collapse for a wide range of VAE models equipped with a concrete theoretical guarantee. We also illustrate the effectiveness of our method through several numerical experiments. |
| title | Controlling Posterior Collapse by an Inverse Lipschitz Constraint on the Decoder Network |
| topic | Machine Learning |
| url | https://arxiv.org/abs/2304.12770 |