Research map / Learning theory and algorithms / Matrix and tensor methods
Large-Scale Sparse Kernel Canonical Correlation Analysis
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 gradKCCA (preimage-constrained KCCA with alternating projected gradient) instead of baselines | non-linear CCA on MNIST · MNIST | test correlation: 0.952 vs 0.950 (KNOI), 0.949 (RCCA) | Helped |
| Use gradKCCA instead of baselines | non-linear CCA on MNIST · MNIST | computation time: 56 s vs 78 s (RCCA, second fastest) and 4578 s (DCCA) | Helped |
| Use gradKCCA instead of baselines | non-linear CCA on MediaMill · MediaMill | test correlation: 0.657 vs 0.645 (KNOI) | Helped |
| Use gradKCCA instead of baselines | non-linear CCA on MediaMill · MediaMill | computation time: 8 s vs 23 s (RCCA, second fastest) | Helped |
| Constrain l1 norm of projection preimages (sparsity) in gradKCCA | gradKCCA on monotone simulated relations · simulated monotone relations with varying noise variables | F1 score and AUC | Helped |
| Use preimage-constrained gradKCCA vs. standard KCCA | non-monotone simulated relations with quadratic kernel · simulated non-monotone relations | test correlation and robustness to noise | No clear effect |
| Use gradKCCA instead of SCCA-HSIC on non-monotone relations | sparse non-linear CCA on simulated non-monotone relations · simulated non-monotone relations | F1 score | Helped |
| Use gradKCCA instead of KCCA/DCCA/KNOI for noise robustness on non-monotone relations | non-linear CCA with irrelevant noise variables · simulated non-monotone relations | robustness to noise variables | Helped |
| Use gradKCCA instead of DCCA/KNOI/SCCA-HSIC on monotone relations with noise | non-linear CCA with irrelevant noise variables · simulated monotone relations | robustness to noise variables | Helped |
| Scale sample size with gradKCCA (linear-time optimization) instead of kernel-matrix methods | large sample simulated monotone relations · simulated monotone relations, n up to 10^6 | running time | Helped |
| Replace KCCA's exact kernel with gradKCCA preimage model on monotone sample-size experiment | gradKCCA vs RCCA and DCCA on monotone relations · simulated monotone relations, varying sample size | train and test kernel canonical correlation | No clear effect |
| Use gradKCCA rather than KNOI/DCCA for non-monotone relations at large sample sizes | non-linear CCA on non-monotone relations · simulated non-monotone relations, varying sample size | test correlation | Mixed |
Closest papers by meaning
- On Sparse Canonical Correlation Analysis (NeurIPS 2024 · similarity 0.92)
- L0-Sparse Canonical Correlation Analysis (ICLR 2022 · similarity 0.89)
- Nonparametric Canonical Correlation Analysis (ICML 2016 · similarity 0.88)
- Unconstrained Stochastic CCA: Unifying Multiview and Self-Supervised Learning (ICLR 2024 · similarity 0.87)
- Unpaired Canonical Correlation Analysis (NeurIPS 2026 · similarity 0.86)
- Efficient Sparse PCA via Block-Diagonalization (ICLR 2025 · similarity 0.83)
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.