Research map / Learning theory and algorithms / Matrix and tensor methods
Streaming Coresets for Symmetric Tensor Factorization
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 online filtering before Lewis-weight sampling (ablation) | Lewis-weight streaming coreset · Streaming tensor coreset | sample size: $d^{p/2}k\epsilon^{-5}$ -> $d^{p/2}k\epsilon^{-5}$ | Mixed |
| Add kernelization to online filtering (ablation) | Online filter · Streaming tensor coreset, odd p | sample size: $n^{1-2/p}dk\epsilon^{-2}$ -> $n^{{(p - 2)}/{(p^2+p)}}d^{p/2+1/4}k^{5/4}\epsilon^{-2}$ | Helped |
| Add kernelization to online filtering (ablation) | Online filter · Streaming tensor coreset, even p | sample size: $n^{1-2/p}dk\epsilon^{-2}$ -> $d^{p/2}k\epsilon^{-2}$ | Helped |
| Use kernelization for odd-order tensor terms (ablation) | Kernel filter · Streaming tensor coreset, odd p | sample size: $n^{1/(p+1)}d^{p/2}k\epsilon^{-2}$ | Helped |
| Add online filtering and Lewis-weight reduction before kernel filtering (ablation) | Kernel filter · Streaming tensor coreset, odd p | sample size: $n^{1/(p+1)}d^{p/2}k\epsilon^{-2}$ -> $d^{p/2+1/2}k^{5/4}\epsilon^{-2}$ | Helped |
| Use matrix Bernstein rather than scalar Bernstein (ablation) | Online row sampling for p=2 · Matrix row sampling | sample size: factor of $O(d^{2})$ -> factor of $O(d\log d)$ | Helped |
Closest papers by meaning
- Streaming Bayesian Deep Tensor Factorization (ICML 2021 · similarity 0.88)
- Tensor Decomposition via Simultaneous Power Iteration (ICML 2017 · similarity 0.87)
- On Tensor Train Rank Minimization : Statistical Efficiency and Scalable Algorithm (NeurIPS 2017 · similarity 0.87)
- Partitioned Tensor Factorizations for Learning Mixed Membership Models (ICML 2017 · similarity 0.86)
- Geometry-aware training of factorized layers in tensor Tucker format (NeurIPS 2024 · similarity 0.86)
- Efficient Leverage Score Sampling for Tensor Train Decomposition (NeurIPS 2024 · 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.