Research map / Learning theory and algorithms / Matrix and tensor methods
A Simple and Fast Algorithm for L1-Norm Kernel PCA
Cheol‐Min Kim, Diego Klabjan
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 Algorithm 1 instead of KPCA-L1 and L1-KPCA on linear kernel | L1-norm kernel PCA · synthetic 1000x50 perturbed data | total explained variation: by 1% and 2% | Improved |
| use L1-norm kernel PCA instead of L2-KPCA | kernel PCA · synthetic perturbed data (linear kernel) | total explained variation: 90% -> 95% | Improved |
| use Algorithm 1 with Gaussian kernel when corrupted fraction r exceeds 20 | kernel PCA · synthetic perturbed data (Gaussian kernel) | total explained variation: 1% to 5% | Improved |
| use R1-KPCA instead of L1-norm methods with linear kernel | kernel PCA · synthetic perturbed data (linear kernel) | total explained variation: 96% vs 95% | Improved |
| use Gaussian kernel instead of linear kernel in Algorithm 1 | Algorithm 1 · Cardio outlier detection | AUC: 0.5790 -> 0.6096 | Improved |
| use Gaussian kernel instead of linear kernel in Algorithm 1 | Algorithm 1 · Ionosphere outlier detection | AUC: 0.6625 -> 0.7238 | Improved |
| use Algorithm 1 Gaussian instead of L2-KPCA Gaussian | outlier detection · Cardio | AUC: 0.4664 -> 0.6096 | Improved |
| use Algorithm 1 instead of iForest on high-dimensional data | outlier detection · Musk | AUC: 0.7596 -> 0.9947 | Improved |
| use iForest instead of Algorithm 1 | outlier detection · WBC | AUC: 0.5292 -> 0.5525 | Hurt |
| use Algorithm 1 instead of KPCA-L1 for runtime | L1-norm kernel PCA · Musk runtime (linear) | runtime (minutes): 69.6 -> 0.5 | Improved |
| use Algorithm 1 instead of KPCA-L1 for runtime | L1-norm kernel PCA · MNIST runtime (linear) | runtime (minutes): 263.3 -> 0.9 | Improved |
| use Algorithm 1 instead of L1-KPCA and R1-KPCA for runtime | robust kernel PCA · MNIST runtime | runtime (minutes): >1440 -> 0.9 | Improved |
Closest papers by meaning
- Extending Kernel PCA through Dualization: Sparsity, Robustness and Fast Algorithms (ICML 2023 · similarity 0.89)
- On the Link Between L1-PCA and ICA (TPAMI 2016 · similarity 0.89)
- Rethinking PCA Through Duality (NeurIPS 2025 · similarity 0.88)
- Bilinear Factor Matrix Norm Minimization for Robust PCA: Algorithms and Applications (TPAMI 2017 · similarity 0.87)
- A Kernel Random Matrix-Based Approach for Sparse PCA (ICLR 2019 · similarity 0.87)
- Efficient Sparse PCA via Block-Diagonalization (ICLR 2025 · 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.
