Reproduce “SimMLM: A Simple Framework for Multi-modal Learning with Missing Modality” on OpenResearch

Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.

Universal app for Apple silicon and Intel. Requires macOS 11 or later.

About this paper

Sijie Li, Chen Chen, Jungong Han

SimMLM combines predictions from available MRI modalities using modality-specific experts and adaptive gating, without synthesizing missing inputs. Its MoFe penalty discourages richer input sets from having higher task error than poorer sets. On BraTS, across 15 settings, it reports mean ET, TC, and WT Dice scores of 67.16, 80.20, and 87.67, higher than MoMKE. The method reduces reversals, not guarantees monotonic accuracy.

Summary by alphaXiv.

Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.