LCIP: Loss-Controlled Inverse Projection of High-Dimensional Image Data
Fuente:
arXiv
Gespeichert in:
| Hauptverfasser: | , , , |
|---|---|
| Format: | Preprint |
| Veröffentlicht: |
2026
|
| Schlagworte: | |
| Online-Zugang: | |
| Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
| _version_ | 1866908828016050176 |
|---|---|
| author | Wang, Yu Dennig, Frederik L. Behrisch, Michael Telea, Alexandru |
| author_facet | Wang, Yu Dennig, Frederik L. Behrisch, Michael Telea, Alexandru |
| contents | Projections (or dimensionality reduction) methods $P$ aim to map high-dimensional data to typically 2D scatterplots for visual exploration. Inverse projection methods $P^{-1}$ aim to map this 2D space to the data space to support tasks such as data augmentation, classifier analysis, and data imputation. Current $P^{-1}$ methods suffer from a fundamental limitation -- they can only generate a fixed surface-like structure in data space, which poorly covers the richness of this space. We address this by a new method that can `sweep' the data space under user control. Our method works generically for any $P$ technique and dataset, is controlled by two intuitive user-set parameters, and is simple to implement. We demonstrate it by an extensive application involving image manipulation for style transfer. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2602_11141 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | LCIP: Loss-Controlled Inverse Projection of High-Dimensional Image Data Wang, Yu Dennig, Frederik L. Behrisch, Michael Telea, Alexandru Human-Computer Interaction Machine Learning Projections (or dimensionality reduction) methods $P$ aim to map high-dimensional data to typically 2D scatterplots for visual exploration. Inverse projection methods $P^{-1}$ aim to map this 2D space to the data space to support tasks such as data augmentation, classifier analysis, and data imputation. Current $P^{-1}$ methods suffer from a fundamental limitation -- they can only generate a fixed surface-like structure in data space, which poorly covers the richness of this space. We address this by a new method that can `sweep' the data space under user control. Our method works generically for any $P$ technique and dataset, is controlled by two intuitive user-set parameters, and is simple to implement. We demonstrate it by an extensive application involving image manipulation for style transfer. |
| title | LCIP: Loss-Controlled Inverse Projection of High-Dimensional Image Data |
| topic | Human-Computer Interaction Machine Learning |
| url | https://arxiv.org/abs/2602.11141 |