Research map / Label-efficient and robust learning / Recognition and long-tail
Deep Partial Multi-View Learning
Changqing Zhang, Yajie Cui, Zongbo Han, Joey Tianyi Zhou, Huazhu Fu, Qinghua Hu
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 |
|---|---|---|---|
| increase missing rate; compare CPM-Nets to LMNN | CPM-Nets · CUB | accuracy: 89.48% vs 86.27% | Improved |
| fill missing views with CRA instead of class averages | compared methods · multi-view datasets (eta=0.5) | accuracy | Improved |
| use CPM-Nets over second-best method | CPM-Nets · CUB, 3Sources-complete, Football, Politics | accuracy: 2.5%, 10.2%, 36.7%, 6.6% | Improved |
| use CPM-Nets on real partial data | CPM-Nets · ADNI and 3Sources-partial | accuracy: 3.4% and 2.3% higher | Improved |
| add label supervision (lambda>0 vs lambda=0) (ablation) | CPM-Nets · Handwritten (eta=0.5) | accuracy | Improved |
| vary trade-off lambda within 0.6-3 at K=128 (ablation) | CPM-Nets · Handwritten (eta=0.5) | accuracy | No clear effect |
| add fine-tuning to reduce train/test label gap (ablation) | CPM-Nets · Handwritten and CUB | accuracy | Improved |
| add adversarial strategy (CPM-GAN vs CPM-without-GAN) (ablation) | CPM-Nets · six datasets, imputation | NRMSE | Improved |
| add adversarial strategy (ablation) | CPM-Nets · Football or CUB, low missing rate, clustering | ACC/NMI | Hurt |
| add adversarial strategy (ablation) | CPM-Nets · clustering, high missing rate | ACC/NMI | Improved |
| use CPM-GAN over second best at eta=0.5 | CPM-GAN · multi-modal datasets clustering | ACC and NMI: 3.36% and 5.11% | Improved |
| use CPM-GAN vs VIGAN (10% complete data for VIGAN) | CPM-GAN · CUB and Animal clustering | ACC and NMI | Improved |
Closest papers by meaning
- CPM-Nets: Cross Partial Multi-View Networks (NeurIPS 2019 · similarity 0.96)
- Multi-View Deep Network for Cross-View Classification (CVPR 2016 · similarity 0.89)
- Rethinking Multi-view Representation Learning via Distilled Disentangling (CVPR 2024 · similarity 0.88)
- Deep Incomplete Multi-view Learning via Cyclic Permutation of VAEs (ICLR 2025 · similarity 0.87)
- Deep Incomplete Multi-View Learning Network with Insufficient Label Information (AAAI 2024 · similarity 0.87)
- Incomplete Contrastive Multi-View Clustering with High-Confidence Guiding (AAAI 2024 · similarity 0.86)
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.
