Research map / Vision-language and multimodal / Cross-modal learning
Structured Multi-modal Graph Disentanglement for Psychiatric Diagnosis
Hongyu Shi, Kaizhong Zheng, Wensheng Zhai, Shuai Jiang, Badong Chen, Liangjun Chen
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 |
|---|---|---|---|
| use SMGD four-subspace factorization instead of the best baseline | SMGD · ABIDE-I | ACC: 0.715 (IBGNN) -> 0.748 | Improved |
| use SMGD instead of the best baseline for cross-dataset transfer | SMGD · ABIDE-I to SRPBS-ASD | ACC: 0.631 (CI-GNN) -> 0.639 | Improved |
| add non-degeneracy constraint (ablation) | SMGD · ABIDE-I | ACC: 0.679 -> 0.748 | Improved |
| add alignment on shared subspace ZS (ablation) | SMGD · ABIDE-I | ACC: 0.671 -> 0.748 | Improved |
| add cross-modal decoupling on complementary subspace ZC (ablation) | SMGD · ABIDE-I | ACC: 0.685 -> 0.748 | Improved |
| add shortcut regularizer Lshort (ablation) | SMGD · ABIDE-I | ACC: 0.709 -> 0.748 | Improved |
| add SMGD factorization on top of basic backbone with fusion (ablation) | GPS backbone with fusion · ABIDE-I | ACC: 0.703 -> 0.748 | Improved |
| replace GW structural alignment with MSE alignment (ablation) | SMGD · ABIDE-I | ACC: 0.748 -> 0.701 | Hurt |
| use InfoNCE alignment to shrink the modality gap (ablation) | Late Fusion / SMGD-CL · ABIDE-I | ACC: modality gap 1.729 but ACC 0.722 vs 0.748 | Hurt |
| use bimodal distortion of learned subspace features (stress test) (ablation) | DDH covariance surrogates · ABIDE-I post-hoc stress test | subspace dependence | Hurt |
Closest papers by meaning
- Multi-modal Medical Diagnosis via Large-small Model Collaboration (CVPR 2025 · similarity 0.85)
- Unified Insights: Harnessing Multi-modal Data for Phenotype Imputation via View Decoupling (NeurIPS 2024 · similarity 0.85)
- Causal Invariance-aware Augmentation for Brain Graph Contrastive Learning (ICML 2025 · similarity 0.84)
- KMD: Koopman Multi-modality Decomposition for Generalized Brain Tumor Segmentation under Incomplete Modalities (CVPR 2025 · similarity 0.84)
- DisenGCD: A Meta Multigraph-assisted Disentangled Graph Learning Framework for Cognitive Diagnosis (NeurIPS 2024 · similarity 0.84)
- Flex-MoE: Modeling Arbitrary Modality Combination via the Flexible Mixture-of-Experts (NeurIPS 2024 · similarity 0.84)
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.
