Show simple item record

dc.contributor.authorGrattarola, D
dc.contributor.authorLivi, L
dc.contributor.authorAlippi, C
dc.date.accessioned2019-07-17T08:58:31Z
dc.date.issued2019-05-22
dc.description.abstractConstant-curvature Riemannian manifolds (CCMs)have been shown to be ideal embedding spaces in many application domains, as their non-Euclidean geometry can naturally account for some relevant properties of data, like hierarchy and circularity. In this work, we introduce the CCM adversarial autoencoder (CCM-AAE), a probabilistic generative model trained to represent a data distribution on a CCM. Our method works by matching the aggregated posterior of the CCM-AAE with a probability distribution defined on a CCM, so that the encoder implicitly learns to represent data on the CCM to fool the discriminator network. The geometric constraint is also explicitly imposed by jointly training the CCM-AAE to maximise the membership degree of the embeddings to the CCM. While a few works in recent literature make use of either hyperspherical or hyperbolic manifolds for different learning tasks, ours is the first unified framework to seamlessly deal with CCMs of different curvatures. We show the effectiveness of our model on three different datasets characterised by non-trivial geometry: semi-supervised classification on MNIST, link prediction on two popular citation datasets, and graph-based molecule generation using the QM9 chemical database. Results show that our method improves upon other autoencoders based on Euclidean and non-Euclidean geometries on all tasks taken into account.en_GB
dc.description.sponsorshipSwiss National Science Foundationen_GB
dc.identifier.citationVol. 81, 105511en_GB
dc.identifier.doi10.1016/j.asoc.2019.105511
dc.identifier.grantnumber200021/172671en_GB
dc.identifier.urihttp://hdl.handle.net/10871/38004
dc.language.isoenen_GB
dc.publisherElsevieren_GB
dc.rights.embargoreasonUnder embargo until 22 May 2020 in compliance with publisher policy.en_GB
dc.rights© 2019 This version is made available under the CC-BY-NC-ND 4.0 license https://creativecommons.org/licenses/by-nc-nd/4.0/ en_GB
dc.subjectAdversarial learningen_GB
dc.subjectConstant-curvature manifoldsen_GB
dc.subjectImage classificationen_GB
dc.subjectLink predictionen_GB
dc.subjectMolecule generationen_GB
dc.titleAdversarial autoencoders with constant-curvature latent manifoldsen_GB
dc.typeArticleen_GB
dc.date.available2019-07-17T08:58:31Z
dc.identifier.issn1568-4946
dc.descriptionThis is the author accepted manuscript. The final version is available from the publisher via the DOI in this recorden_GB
dc.identifier.journalApplied Soft Computingen_GB
dc.rights.urihttps://creativecommons.org/licenses/by-nc-nd/4.0/ en_GB
dcterms.dateAccepted2019-05-18
rioxxterms.versionAMen_GB
rioxxterms.licenseref.startdate2019-05-18
rioxxterms.typeJournal Article/Reviewen_GB
refterms.dateFCD2019-07-17T08:55:23Z
refterms.versionFCDAM
refterms.panelBen_GB


Files in this item

This item appears in the following Collection(s)

Show simple item record

© 2019  This version is made available under the CC-BY-NC-ND 4.0 license https://creativecommons.org/licenses/by-nc-nd/4.0/ 
Except where otherwise noted, this item's licence is described as © 2019 This version is made available under the CC-BY-NC-ND 4.0 license https://creativecommons.org/licenses/by-nc-nd/4.0/