Research map / Optimization / Optimizers
Efficient Globally Convergent Stochastic Optimization for 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 stochastic variance reduction to solve the least-squares subproblems instead of batch gradient-based methods | ALS-VR and SI-VR · Mediamill, JW11, and MNIST | objective suboptimality versus data passes | Helped |
| Use shift-and-invert preconditioning rather than ALS | SVRG-based CCA · Mediamill, JW11, and MNIST | convergence speed | Helped |
| Use ASVRG rather than SVRG for the least-squares subproblems | ALS and SI · Mediamill, JW11, and MNIST with large condition numbers | convergence speed | Helped |
| Use inexact minibatch covariance estimates for normalization rather than exact normalization | stochastic AppGrad · Mediamill, JW11, and MNIST | objective suboptimality | Hurt |
| Use shift-and-invert preconditioning rather than ALS | CCA optimization · CCA time complexity | time complexity: $\tilde{\calO} \left( d (N + (\tilde{\kappa} \frac{1}{\rho_1 - \rho_2})^2 ) \cdot \log^2 \left(\frac{1}{\eta}\right) \ri | Helped |
| Increase regularization to improve least-squares conditioning | ALS and SI · Mediamill, JW11, and MNIST | least-squares condition number | Helped |
Closest papers by meaning
- Stochastic Approximation for Canonical Correlation Analysis (NeurIPS 2017 · similarity 0.92)
- Unconstrained Stochastic CCA: Unifying Multiview and Self-Supervised Learning (ICLR 2024 · similarity 0.90)
- An Online Riemannian PCA for Stochastic Canonical Correlation Analysis (NeurIPS 2021 · similarity 0.89)
- On Sparse Canonical Correlation Analysis (NeurIPS 2024 · similarity 0.89)
- Efficient Algorithms for Large-scale Generalized Eigenvector Computation and Canonical Correlation Analysis (ICML 2016 · similarity 0.89)
- Unpaired Canonical Correlation Analysis (NeurIPS 2026 · 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.