Research map / Learning theory and algorithms / Clustering
OPTION: Optimal Transport–Guided Flow Matching for Incomplete and Unaligned Multi-View Clustering
Siyuan Zhou, Zhibin Gu
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 instance-level contrastive loss L_con (ablation) | OPTION · Scene-15 (aligned, 70% missing) | ACC: 25.3 -> 37.5 | Improved |
| add DEC clustering loss L_clu (ablation) | OPTION · Scene-15 (aligned, 70% missing) | ACC: 26.5 -> 37.5 | Improved |
| add reconstruction loss L_rec (ablation) | OPTION · Scene-15 (aligned, 70% missing) | ACC: 32.4 -> 37.5 | Improved |
| add GW-inspired structural loss L_gw (ablation) | OPTION · Scene-15 (aligned, 70% missing) | ACC: 36.4 -> 37.5 | Improved |
| add conditional flow matching imputation L_flow (ablation) | OPTION · Scene-15 (aligned, 70% missing) | ACC: 36.3 -> 37.5 | Improved |
| replace 100-step diffusion (DCG) with 10-step flow matching | OPTION · Six-dataset average inference speed | inference time (ms/batch): 25.95 -> 3.28 | Improved |
| use flow matching with GW alignment instead of diffusion DCG at high missingness | OPTION · CUB (70% missing) | ACC: 18.2 -> 74.7 | Improved |
| vary ODE steps from 10 to 50 (ablation) | OPTION · Scene-15 sensitivity | ACC | No clear effect |
| vary lambda_g over two orders of magnitude (ablation) | OPTION · Scene-15 sensitivity | ACC: S = 0.082 | No clear effect |
| increase unaligned rate from 0% to 60% for OPTION | OPTION · BDGP unaligned | ACC: 50.3 -> 46.3 | Hurt |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 0 of 1 on most shared results.
- PROTOCOL — PROTOCOL: Partial Optimal Transport-enhanced Contrastive Learning for Imbalanced Multi-view Clustering (ICML 2025) · better on 0 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
- Straight-Path Flow Matching for Incomplete Multi-View Clustering (ECCV 2026 · similarity 0.92)
- Transport Barycenter-Guided Sample-to-Cluster Matching for Unaligned Multi-view Clustering (TPAMI 2026 · similarity 0.92)
- Deep Incomplete Multi-View Clustering With Cross-View Partial Sample and Prototype Alignment (CVPR 2023 · similarity 0.90)
- Robust Multi-view Clustering against Imperfect Information (NeurIPS 2026 · similarity 0.88)
- Medusa: A Multi-Scale High-order Contrastive Dual-Diffusion Approach for Multi-View Clustering (CVPR 2025 · similarity 0.88)
- Incomplete Contrastive Multi-View Clustering with High-Confidence Guiding (AAAI 2024 · similarity 0.88)
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.
