Publication:
Cherry-Picking Gradients: Learning Low-Rank Embeddings of Visual Data via Differentiable Cross-Approximation

Simple item page
dc.contributor.authorBallester, Rafael
dc.contributor.authorUsvyatsov, Mikhail
dc.contributor.authorMakarova, Anastasia
dc.contributor.authorRakhuba, Maxim
dc.contributor.authorKrause, Andreas
dc.contributor.authorSchindler, Konrad
dc.contributor.rorhttps://ror.org/02jjdwm75
dc.date.accessioned2025-04-03T15:50:24Z
dc.date.available2025-04-03T15:50:24Z
dc.date.issued2021-11-15
dc.description.abstractWe propose an end-to-end trainable framework that processes large-scale visual data tensors by looking at a fraction of their entries only. Our method combines a neural network encoder with a tensor train decomposition to learn a low-rank latent encoding, coupled with cross-approximation (CA) to learn the representation through a subset of the original samples. CA is an adaptive sampling algorithm that is native to tensor decompositions and avoids working with the full high-resolution data explicitly. Instead, it actively selects local representative samples that we fetch out-of-core and on-demand. The required number of samples grows only logarithmically with the size of the input. Our implicit representation of the tensor in the network enables processing large grids that could not be otherwise tractable in their uncompressed form. The proposed approach is particularly useful for large-scale multidimensional grid data (e.g., 3D tomography), and for tasks that require context over a large receptive field (e.g., predicting the medical condition of entire organs).
dc.description.peerreviewedyes
dc.description.statusPublished
dc.formatapplication/pdf
dc.identifier.citationUsvyatsov, M., Makarova, A., Ballester-Ripoll, R., Rakhuba, M., Krause, A., & Schindler, K. (2021). Cherry-picking gradients: Learning low-rank embeddings of visual data via differentiable cross-approximation. In Proceedings of the IEEE/CVF International Conference on Computer Vision (pp. 11426-11435). https://doi.org/10.48550/arXiv.2105.14250.
dc.identifier.doihttps://doi.org/10.48550/arXiv.2105.14250
dc.identifier.issn1550-5499
dc.identifier.urihttps://hdl.handle.net/20.500.14417/3700
dc.journal.titleProceedings of the IEEE/CVF International Conference on Computer Vision
dc.language.isoen
dc.page.final11435
dc.page.initial11426
dc.page.total16
dc.publisherCornell University
dc.relation.departmentApplied Mathematics
dc.relation.entityIE University
dc.relation.schoolIE School of Science & Technology
dc.rightsAttribution 4.0 International
dc.rights.accessRightsinfo:eu-repo/semantics/openAccess
dc.rights.urihttps://creativecommons.org/licenses/by/4.0/deed
dc.titleCherry-Picking Gradients: Learning Low-Rank Embeddings of Visual Data via Differentiable Cross-Approximation
dc.typeinfo:eu-repo/semantics/article
dc.version.typeinfo:eu-repo/semantics/publishedVersion
dspace.entity.typePublication
relation.isAuthorOfPublication6f756541-9eb4-430c-9664-1833c080ce57
relation.isAuthorOfPublication.latestForDiscovery6f756541-9eb4-430c-9664-1833c080ce57
Files
Original bundle
Now showing 1 - 1 of 1
Thumbnail Image
Name:
CHERRY-PICKING GRADIENTS LEARNING LOW-RANK.pdf
Size:
25.06 MB
Format:
Adobe Portable Document Format
Download
License bundle
Now showing 1 - 1 of 1
No Thumbnail Available
Name:
license.txt
Size:
2.83 KB
Format:
Item-specific license agreed to upon submission
Description:
Download
Collections
Research Articles