LARGO: Low-Rank Hypernetwork for Handling Missing Modalities

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Vyncke, Niels, Ashtari, Pooya, Pižurica, Aleksandra
Format: Preprint
Published: 2026
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866917467859714048
author Vyncke, Niels
Ashtari, Pooya
Pižurica, Aleksandra
author_facet Vyncke, Niels
Ashtari, Pooya
Pižurica, Aleksandra
contents Addressing missing modalities is an important challenge in multimodal image analysis and often relies on complex architectures that do not transfer easily to different datasets without architectural modifications or hyperparameter tuning. While most existing methods tackle this problem in feature space by engineering representations that are robust to missing inputs, we instead operate in weight space. We propose LARGO, a hypernetwork that compresses the $2^N-1$ dedicated missing-modality models into a single network by modelling the convolutional weights using the Canonical Polyadic (CP) tensor decomposition. Extensive experimental validation on BraTS 2018 (4 modalities, 15 scenarios) and ISLES 2022 (3 modalities, 7 scenarios) shows that our method ranks first in 47 out of 52 configurations, achieving average Dice improvements of +0.68$\%$ and +2.53$\%$ over state-of-the-art baselines (mmFormer, M$^{3}$AE, ShaSpec, SimMLM). A proof-of-concept experiment on avMNIST suggests that LARGO may extend beyond medical imaging to heterogeneous non-medical modalities.
format Preprint
id arxiv_https___arxiv_org_abs_2605_06086
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle LARGO: Low-Rank Hypernetwork for Handling Missing Modalities
Vyncke, Niels
Ashtari, Pooya
Pižurica, Aleksandra
Computer Vision and Pattern Recognition
Addressing missing modalities is an important challenge in multimodal image analysis and often relies on complex architectures that do not transfer easily to different datasets without architectural modifications or hyperparameter tuning. While most existing methods tackle this problem in feature space by engineering representations that are robust to missing inputs, we instead operate in weight space. We propose LARGO, a hypernetwork that compresses the $2^N-1$ dedicated missing-modality models into a single network by modelling the convolutional weights using the Canonical Polyadic (CP) tensor decomposition. Extensive experimental validation on BraTS 2018 (4 modalities, 15 scenarios) and ISLES 2022 (3 modalities, 7 scenarios) shows that our method ranks first in 47 out of 52 configurations, achieving average Dice improvements of +0.68$\%$ and +2.53$\%$ over state-of-the-art baselines (mmFormer, M$^{3}$AE, ShaSpec, SimMLM). A proof-of-concept experiment on avMNIST suggests that LARGO may extend beyond medical imaging to heterogeneous non-medical modalities.
title LARGO: Low-Rank Hypernetwork for Handling Missing Modalities
topic Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2605.06086