Transfer learning RGB models to hyperspectral images with trainable tensor decompositions

Mariette Schönfeld, Laurens Devos, Wannes Meert, Hendrik Blockeel · 2026

Read Transfer learning RGB models to hyperspectral images with trainable tensor decompositions by Mariette Schönfeld, Laurens Devos, Wannes Meert, Hendrik Blockeel (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2605-28331.txt.

Description

Schönfeld and colleagues carry RGB-trained vision networks over to hyperspectral images without discarding image or model information. Partially trainable tensor decompositions split pretrained filters into spatial and spectral parts, replacing only the spectral parts with higher-channel trainable versions. On varied hyperspectral datasets the method proved more accurate and robust than rival transfer approaches.

Catalog record

CREATORMariette Schönfeld, Laurens Devos, Wannes Meert, Hendrik Blockeel
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE36525
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2605.28331

Browse the full library · About the Tower of Babel