Research map / Vision-language and multimodal / Cross-modal learning
Residual-Guided Expert Specialization for Incomplete Multimodal Learning
Seunghun Baek, Jihwan Park, Jaeyoon Sim, Minjae Jeong, Hoseok Lee, Won Hwa Kim
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 changed | Where | Measured | Effect |
|---|---|---|---|
| add a mixture-of-experts structure (ablation) | ResNet-18 · CASIA-SURF | ACER (%): 7.52 -> 4.12 | Improved |
| add residual routing (ablation) | MoE · CASIA-SURF | ACER (%): 4.12 -> 3.78 | Improved |
| add noise regularization (ablation) | MoE with residual routing · CASIA-SURF | ACER (%): 3.78 -> 2.72 | Improved |
| add discrepancy-guided sampling (ablation) | MoE with residual routing and noise · CASIA-SURF | ACER (%): 2.72 -> 2.43 | Improved |
| add MoE (ablation) | DeepLab v3+ · MCubeS | mIoU: 0.4257 -> 0.4327 | Improved |
| add residual routing (ablation) | MoE · MCubeS | mIoU: 0.4327 -> 0.4661 | Improved |
| add noise regularization (ablation) | MoE with residual routing · MCubeS | mIoU: 0.4661 -> 0.4750 | Improved |
| add discrepancy-guided sampling (ablation) | MoE with residual routing and noise · MCubeS | mIoU: 0.4750 -> 0.4773 | Improved |
| use residual as router input instead of concatenation (ablation) | MARS · CASIA-SURF | ACER (%): 3.83 -> 2.43 | Improved |
| use residual as router input instead of attention fusion (ablation) | MARS · CASIA-SURF | ACER (%): 4.83 -> 2.43 | Improved |
| use residual as router input instead of ratio (ablation) | MARS · CASIA-SURF | ACER (%): 12.01 -> 2.43 | Improved |
| feed true full-modality residuals at test time (oracle) (ablation) | MARS · CASIA-SURF | ACER (%): 2.43 -> 0.94 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 2 of 2 on most shared results.
- SimMLM — SimMLM: A Simple Framework for Multi-modal Learning with Missing Modality (ICCV 2025) · better on 1 of 1 shared result
- Flex-MoE — Flex-MoE: Modeling Arbitrary Modality Combination via the Flexible Mixture-of-Experts (NeurIPS 2024) · better on 1 of 1 shared result
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
- MMANet: Margin-Aware Distillation and Modality-Aware Regularization for Incomplete Multimodal Learning (CVPR 2023 · similarity 0.88)
- R2-T2: Re-Routing in Test-Time for Multimodal Mixture-of-Experts (ICML 2025 · similarity 0.87)
- Flex-MoE: Modeling Arbitrary Modality Combination via the Flexible Mixture-of-Experts (NeurIPS 2024 · similarity 0.86)
- Do More Modalities Always Help? A Geometric Perspective on Missing-Modality Robustness (NeurIPS 2026 · similarity 0.86)
- Knowledge Bridger: Towards Training-Free Missing Modality Completion (CVPR 2025 · similarity 0.85)
- Unity by Diversity: Improved Representation Learning for Multimodal VAEs (NeurIPS 2024 · similarity 0.85)
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.
