AI Paper Atlas

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

SOTAlign: Semi-Supervised Alignment of Unimodal Vision and Language Models via Optimal Transport

Simon Roschmann, Paul Krzakala, Sonia Mazelet, Quentin Bouniot, Zeynep Akata

ICML 2026 · Vision-language and multimodal › Cross-modal learning · approach: Contrastive vision-language alignment · 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
replace CKA divergence with KLOT in the regularizer (ablation)SOTAlign with CCA teacher · COCO MeanR@1MeanR@1: 23.5 -> 30.3Improved
replace InfoNCE divergence with KLOT in the regularizer (ablation)SOTAlign with CCA teacher · COCO MeanR@1MeanR@1: 24.1 -> 30.3Improved
add unpaired-data KLOT regularization to a CCA linear teacher (ablation)CCA alignment · COCO MeanR@1MeanR@1: 21.5 -> 30.3Improved
use CCA teacher instead of contrastive linear teacher with KLOT (ablation)SOTAlign · COCO MeanR@1MeanR@1: 28.5 -> 30.3Improved
use CKA with a contrastive teacher (ablation)SOTAlign · COCO MeanR@1MeanR@1: 24.2 -> 24.2No clear effect
add unpaired images and text via SOTAlign vs supervised SAIL10k-pair alignment · ImageNet-1k zero-shottop-1 accuracy: 35.6 -> 46.1Improved
add unpaired samples as extra negatives or pseudo-labels (semi-supervised SAIL, NNCLR, S-CLIP)supervised baselines · COCO retrievalT2I Recall@1: 21.0 -> 20.7 (SAIL)No clear effect
increase number of paired samples (ablation)SOTAlign vs SAIL · COCO MeanR@1MeanR@1: gains up to +10% between 10^3 and 10^4 pairsMixed
use out-of-distribution unpaired data (CC12M, COCO, ImageNet-1k, synthetic captions) (ablation)SOTAlign · cross-dataset unpaired dataretrieval and classification: Pearson r = -0.76 with shiftImproved
replace unpaired data with Gaussian-sampled embeddings (ablation)SOTAlign · COCO retrievalretrievalNo clear effect
use CC3M pairs with synthetic captions (ablation)SOTAlign · COCO T2I R@1Recall@1: 26.5 -> 31.3Improved
use DINOv3 instead of DINOv2 vision encoder with NV-Embed-v2 (ablation)SOTAlign · ImageNet-1k zero-shottop-1 accuracy: 42.5 -> 46.1Improved

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.

Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.

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