AI Paper Atlas

Research map / Label-efficient and robust learning / Recognition and long-tail

Multi-Metric Representation Learning Strategy Based on Clustering for Fine-Grained Multimodal Sentiment Analysis

Yidan Wang, Zongheng Wang, Hongjie Xing, Chunguo Li, Xiaoxiao Liu

CVPR 2026 · Label-efficient and robust learning › Recognition and long-tail · approach: Multilingual language representation pretraining

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
add MMC loss (multi-metric clustering loss) (ablation)MMRest · CMU-MOSIAcc-5: 51.60 -> 54.08Improved
add geometric projection bias from learned metrics (ablation)MMRest · CMU-MOSIAcc-5: 54.08 -> 56.42Improved
add MMC loss (ablation)MMRest · CMU-MOSEIAcc-7: 53.74 -> 53.98Improved
add projection bias (ablation)MMRest · CMU-MOSEIAcc-7: 53.98 -> 54.72Improved
use learned multi-metric matrices instead of a single identity metric (ablation)MMRest · CMU-MOSIAcc-7: 32.94 -> 48.98Improved
use learned multi-metric matrices instead of a single identity metric (ablation)MMRest · CMU-MOSIMAE: 0.914 -> 0.683Improved
use audio-only or visual-only input (ablation)MMRest · CMU-MOSIAcc-5: 15.16 / 18.51 vs 56.42 fullHurt
use multi-metric clustering instead of cross-modal attention fusionMMRest vs MCL-MCF · CMU-MOSIParameters: 199.30M -> 0.37MImproved
MMRest vs MCL-MCF on CMU-MOSEI Acc-2 and CorrMMRest · CMU-MOSEICorr: 0.769 vs DEVA 0.769No clear effect

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