Research map / Continual learning / Online and replay-based
Towards Multimodal Continual Knowledge Embedding with Modality Forgetting Modulation
Xiaowen Jiang, Jing Yang, ShunDong Yang, Yuan Gao, Xinfa Jiang, Laurence Tianruo Yang, Jieming Yang
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 historical path preservation (ablation) | MoFot · M-RELATION | MRR: 0.148 -> 0.338 | Improved |
| add historical path preservation (ablation) | MoFot · M-FACT | BWT: -0.0826 -> -0.0180 | Improved |
| add path-guided multimodal feature modulation (ablation) | MoFot · M-HYBRID | MRR: 0.312 -> 0.353 | Improved |
| add weight interpolation update (ablation) | MoFot · M-HYBRID | MRR: 0.333 -> 0.353 | Improved |
| add weight interpolation update (ablation) | MoFot · M-RELATION | BWT: -0.0386 -> -0.0121 | Improved |
| naively add multimodal features to unimodal continual models (LKGE-M, IncDE-M) | MoFot · M-ENTITY | MRR: LKGE 0.234 -> 0.213; IncDE 0.253 -> 0.220 | Hurt |
| remove PGF and WIU from multimodal MoFot (ablation) | MoFot · M-RELATION | forgetting of snapshot 1: unimodal forgets less than multimodal without PGF/WIU | Hurt |
| add textual and visual modalities (ablation) | MoFot · M-RELATION | MRR/Hits: Figure 6(b), no numbers | Improved |
| use MoFot vs IndMKG on M-RELATION Hits@10 | MoFot · M-RELATION | Hits@10: 0.525 -> 0.520 | No clear effect |
| use MoFot instead of fine-tuned static MMKG models (IMF, MoCi) | MoFot · M-ENTITY | MRR: 0.221 -> 0.413 | Improved |
| use MoFot instead of best unimodal continual model | MoFot · M-ENTITY | BWT: -0.006 -> 0.011 | Improved |
Closest papers by meaning
- Towards Dynamic Modality Alignment in Multimodal Continual Learning (CVPR 2026 · similarity 0.89)
- Mitigating Intra- and Inter-modal Forgetting in Continual Learning of Unified Multimodal Models (NeurIPS 2025 · similarity 0.87)
- Knowledge Graph Enhanced Generative Multi-modal Models for Class-Incremental Learning (NeurIPS 2025 · similarity 0.86)
- FLAME: Adaptive Mixture-of-Experts for Continual Multimodal Multi-Task Learning (NeurIPS 2026 · similarity 0.85)
- CMT: A Memory Compression Method for Continual Knowledge Learning of Large Language Models (AAAI 2025 · similarity 0.85)
- Continual Multimodal Contrastive Learning (NeurIPS 2025 · 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.
