AI Paper Atlas

Research map / Vision-language and multimodal / Cross-modal learning

Synesthesia via Direct Latent Augmentation: Bypassing the Decode-Encode Loop for Cross-Modal Distillation

Cristian Sbrolli, Nicolas Michel, Matteo Matteucci, Toshihiko Yamasaki

ECCV 2026 · Vision-language and multimodal › Cross-modal learning · approach: Unimodal-to-multimodal representation transfer · 0 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
train teacher on undecoded latents instead of decoded raw synthetic datamultimodal teacher · Caltech101Top-1 accuracy: +7.39%Improved
add synthetic latents (3D + audio) to image (ablation)Image-only teacher · Caltech101Top-1 accuracy: 55.10 -> 66.47Improved
add 3D latents to image (ablation)Image-only teacher · ImageNet-100Top-1 accuracy: 61.57 -> 70.03Improved
add audio latents to image (ablation)Image-only teacher · ImageNet-100Top-1 accuracy: 61.57 -> 65.76Improved
add audio to Image+3D (full three modalities) (ablation)Image + 3D teacher · ImageNet-100Top-1 accuracy: 70.03 -> 71.59Improved
use Perceiver resampler for raw decoded modalities (ablation)Raw teacher · multiple benchmarksTop-1 accuracy: at most 1.13 ppImproved
use MLP projection head instead of L2 matching (CLS target) (ablation)LUGPI student · ImageNet-100Top-1 accuracy: 65.70 -> 67.04Improved
predict auxiliary-modality summaries instead of teacher CLS (ablation)CLS-Pred student · ImageNet-100Top-1 accuracy: 67.04 -> 68.54Improved
use DLA latent teacher instead of raw teacher (ablation)LUGPI student · ImageNet-100Top-1 accuracy: 62.31 -> 65.70Improved
use DLA latent teacher instead of raw teacher (ablation)MESSy student · ImageNet-100Top-1 accuracy: 64.03 -> 68.54Improved
use MESSy instead of LUGPI distillationstudent · Caltech101Top-1 accuracy: +3.40%Improved
use MESSy instead of LUGPI distillationstudent · CIFAR-100 linear probeaccuracy: 42.4 -> 48.1Improved

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy