Research map / Learning theory and algorithms / Matrix and tensor methods
An Online Riemannian PCA for Stochastic 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 RSG+ instead of MSG for streaming top-k CCA | CCA · MNIST, Mediamill, and CIFAR-10 | PCC | Helped |
| Use RSG+ instead of MSG when the leading covariance eigenvalue dominates and k > 1 | CCA · Mediamill, k=2 and k=4 | PCC | Hurt |
| Train with RSG+ rather than batch DeepCCA | DeepCCA feature learning · Noisy MNIST, d=100 | Accuracy(%): 80.57 -> 79.79 | Hurt |
| Use RSG+ to train DeepCCA features at d=500 and d=1000 | DeepCCA feature learning · Noisy MNIST, d=500 and d=1000 | Accuracy(%) | Helped |
| Add CCA loss at conv0 | CelebA attractiveness classifier · CelebA | DEO(%): 22.3 -> 17.4 | Helped |
| Add CCA loss at conv0 | CelebA attractiveness classifier · CelebA | DDP(%): 4.8 -> 1.4 | Helped |
| Add CCA loss at conv1 | CelebA attractiveness classifier · CelebA | DEO(%): 22.3 -> 15.3 | Helped |
| Add CCA loss at conv1 | CelebA attractiveness classifier · CelebA | Accuracy(%): 76.3 -> 77.7 | Helped |
| Add CCA loss at conv2 | CelebA attractiveness classifier · CelebA | DEO(%): 22.3 -> 22.0 | No clear effect |
| Add CCA loss at conv2 | CelebA attractiveness classifier · CelebA | DDP(%): 4.8 -> 2.8 | Helped |
| Apply CCA loss to stacked features from different layers | CelebA attractiveness classifier · CelebA | DEO(%): 22.3 -> 22.1 | No clear effect |
| Apply CCA loss to stacked features from different layers | CelebA attractiveness classifier · CelebA | DDP(%): 4.8 -> 3.9 | Helped |
Closest papers by meaning
- Intrinsic Riemannian Cross-covariance for Manifold-valued Random Objects (NeurIPS 2026 · similarity 0.87)
- Unconstrained Stochastic CCA: Unifying Multiview and Self-Supervised Learning (ICLR 2024 · similarity 0.86)
- SERENA: A Unified Stochastic Recursive Variance Reduced Gradient Framework for Riemannian Non-Convex Optimization (ICML 2025 · similarity 0.86)
- Optimization without Retraction on the Random Generalized Stiefel Manifold (ICML 2024 · similarity 0.86)
- Efficient Optimization with Orthogonality Constraint: a Randomized Riemannian Submanifold Method (ICML 2025 · similarity 0.85)
- Decentralized Projected Riemannian Stochastic Recursive Momentum Method for Nonconvex Optimization (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.