Research map / Graph learning / Graph algorithms and clustering
Provable Self-Representation Based Outlier Detection in a Union of Subspaces
Chong You, Daniel P. Robinson, Rene Vidal
Check 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 R-graph instead of Outlier Pursuit | Outlier detection · Extended Yale B, one subject, 35% outliers | AUC: 0.972 -> 0.986 | Improved |
| Use R-graph instead of Outlier Pursuit | Outlier detection · Extended Yale B, three subjects, 15% outliers | AUC: 0.968 -> 0.985 | Improved |
| Use R-graph instead of LRR | Outlier detection · Caltech-256, one category, 50% outliers | AUC: 0.907 -> 0.948 | Improved |
| Use R-graph instead of L1-thresholding | Outlier detection · Caltech-256, three categories, 50% outliers | AUC: 0.810 -> 0.929 | Improved |
| Use R-graph instead of L1-thresholding | Outlier detection · Caltech-256, five categories, 50% outliers | AUC: 0.774 -> 0.913 | Improved |
| Use R-graph instead of L1-thresholding | Outlier detection · Coil-100, one category, 50% outliers | AUC: 0.991 -> 0.997 | Improved |
| Use R-graph instead of L1-thresholding | Outlier detection · Coil-100, four categories, 25% outliers | F1: 0.941 -> 0.970 | Improved |
| Use R-graph instead of L1-thresholding | Outlier detection · Coil-100, seven categories, 15% outliers | F1: 0.897 -> 0.955 | Improved |
| Use self-representation connection patterns rather than representation-vector magnitude | Outlier detection on Extended Yale B face images · Extended Yale B | outlier detection performance | Improved |
| Increase the number of inlier categories | OutRank and CoP · Caltech-256 | AUC | Hurt |
| Vary alpha over the tested range [1, 50] | R-graph · Extended Yale B, three inlier groups and 15% outliers | performance | No clear effect |
| Vary the percentage of outliers from 1% to 15% | R-graph · Extended Yale B, three inlier groups | performance | No clear effect |
Closest papers by meaning
- Latent Discriminant Subspace Representations for Multi-View Outlier Detection (AAAI 2018 · similarity 0.88)
- Outlier Detection for Robust Multi-Dimensional Scaling (TPAMI 2018 · similarity 0.87)
- Escaping the Homophily Trap: A Threshold-free Graph Outlier Detection Framework via Clustering-guided Edge Reweighting (ICLR 2026 · similarity 0.87)
- Accelerated Combinatorial Search for Outlier Detection with Provable Bound on Sub-Optimality (AAAI 2021 · similarity 0.86)
- Quantile-Based Maximum Likelihood Training for Outlier Detection (AAAI 2024 · similarity 0.85)
- Multi-Subspace Matrix Recovery from Permuted Data (AAAI 2025 · similarity 0.85)
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.
