Fast and Accurate Tensor Completion with Total Variation Regularized Tensor Trains
We propose a new tensor completion method based on tensor trains. The to-be-completed tensor is modeled as a low-rank tensor train, where we use the known tensor entries and their coordinates to update the tensor train. A novel tensor train initialization procedure is proposed specifically for image...
| Main Authors: | , , , |
|---|---|
| Other Authors: | , |
| Format: | Article |
| Language: | English |
| Published: |
Institute of Electrical and Electronics Engineers (IEEE),
2021-03-05T12:33:10Z.
|
| Subjects: | |
| Online Access: | Get fulltext |