Deep probabilistic generative models have achieved incredible success in many fields of application. Among such models, variational autoencoders (VAEs) have proved their ability in modeling a generative process by learning a latent representation of the input. In this paper, we propose a novel VAE defined in the quaternion domain, which exploits the properties of quaternion algebra to improve performance while significantly reducing the number of parameters required by the network. The success of the proposed quaternion VAE with respect to traditional VAEs relies on the ability to leverage the internal relations between quaternion-valued input features and on the properties of second-order statistics which allow to define the latent variables in the augmented quaternion domain. In order to show the advantages due to such properties, we define a plain convolutionalVAE in the quaternion domain and we evaluate its performance with respect to its real-valued counterpart on the CelebA face dataset.
A quaternion-valued variational autoencoder / Grassucci, E.; Comminiello, D.; Uncini, A.. - 2021:(2021), pp. 3310-3314. (Intervento presentato al convegno 2021 IEEE International Conference on Acoustics, Speech, and Signal Processing, ICASSP 2021 tenutosi a Toronto; Canada) [10.1109/ICASSP39728.2021.9413859].
A quaternion-valued variational autoencoder
Grassucci E.
;Comminiello D.;Uncini A.
2021
Abstract
Deep probabilistic generative models have achieved incredible success in many fields of application. Among such models, variational autoencoders (VAEs) have proved their ability in modeling a generative process by learning a latent representation of the input. In this paper, we propose a novel VAE defined in the quaternion domain, which exploits the properties of quaternion algebra to improve performance while significantly reducing the number of parameters required by the network. The success of the proposed quaternion VAE with respect to traditional VAEs relies on the ability to leverage the internal relations between quaternion-valued input features and on the properties of second-order statistics which allow to define the latent variables in the augmented quaternion domain. In order to show the advantages due to such properties, we define a plain convolutionalVAE in the quaternion domain and we evaluate its performance with respect to its real-valued counterpart on the CelebA face dataset.File | Dimensione | Formato | |
---|---|---|---|
Grassucci_Quaternion-valued_2021.pdf
accesso aperto
Tipologia:
Versione editoriale (versione pubblicata con il layout dell'editore)
Licenza:
Tutti i diritti riservati (All rights reserved)
Dimensione
3.79 MB
Formato
Adobe PDF
|
3.79 MB | Adobe PDF |
I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.