Research map / Label-efficient and robust learning / Recognition and long-tail
EXOTIC: External Vision-driven Incomplete Multi-view Classification
Shilin Xu, Dezhong Peng, Zhenwen Ren, Yuan Sun
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 external knowledge completion of missing views (ablation) | EXOTIC · Caltech101 (missing rate 0.9) | accuracy: 85.38 -> 88.16 | Improved |
| add external knowledge completion of missing views (ablation) | EXOTIC · NUSWIDE (missing rate 0.9) | accuracy: 39.06 -> 43.90 | Improved |
| add external knowledge completion of missing views (ablation) | EXOTIC · Scene15 (missing rate 0.9) | accuracy: 64.72 -> 62.53 | Hurt |
| add category alignment loss Lvc (ablation) | EXOTIC · LandUse21 (missing rate 0.9) | accuracy: 51.05 -> 55.24 | Improved |
| add category alignment loss Lvc (ablation) | EXOTIC · Caltech101 (missing rate 0.9) | accuracy: 84.31 -> 88.16 | Improved |
| add knowledge alignment loss Lkc (ablation) | EXOTIC · Caltech101 (missing rate 0.9) | accuracy: 85.47 -> 88.16 | Improved |
| add knowledge alignment loss Lkc (ablation) | EXOTIC · NUSWIDE (missing rate 0.9) | accuracy: 44.12 -> 43.90 | Hurt |
| use a small 100-image external library instead of larger ones (ablation) | EXOTIC · HW (missing rate 0.9) | accuracy: 94.20 -> 59.25 | Hurt |
| increase external library size from 500 to 50,000 ImageNet images (ablation) | EXOTIC · Caltech101 (missing rate 0.9) | accuracy: 87.41 -> 88.16 | No clear effect |
| use cross-domain cartoon library instead of natural images (ablation) | EXOTIC · LandUse21 (missing rate 0.9) | accuracy: 55.52 vs 55.24 | No clear effect |
| increase alpha and beta too much (ablation) | EXOTIC · HW | accuracy | Hurt |
| use external knowledge completion versus zero or mean filling (ablation) | EXOTIC · HW (missing rate 0.9) | t-SNE class separation | Improved |
Closest papers by meaning
- Learning Compact Semantic Information for Incomplete Multi-View Missing Multi-Label Classification (ICML 2025 · similarity 0.87)
- Deep Incomplete Multi-view Learning via Cyclic Permutation of VAEs (ICLR 2025 · similarity 0.85)
- Global-Semantic Alignment Distillation for Partial Multi-view Classification (AAAI 2025 · similarity 0.85)
- Incomplete Contrastive Multi-View Clustering with High-Confidence Guiding (AAAI 2024 · similarity 0.85)
- Incomplete Multi-view Clustering via Hierarchical Semantic Alignment and Cooperative Completion (NeurIPS 2025 · similarity 0.84)
- Deep Incomplete Multi-View Learning Network with Insufficient Label Information (AAAI 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.
