Research map / Learning theory and algorithms / Sample complexity
TIMERS: Error-Bounded SVD Restart on Dynamic Networks
Ziwei Zhang, Peng Cui, Jian Pei, Xiao Wang, Wenwu Zhu
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 TIMERS rather than Heu-FL for restart timing at the same restart count | SVD on dynamic networks · Five real networks | maximum relative reconstruction error: 27% to 42% reduction | Improved |
| Use TIMERS rather than existing methods at the same maximum tolerated error | SVD on dynamic networks · Real dynamic networks | number of restarts: 25% to 50% reduction | Improved |
| Monitor the error margin using a lower bound on minimum SVD loss to trigger restarts | TIMERS · FACEBOOK link prediction | relative error of MSE (%): 4.27 -> 1.54 | Improved |
| Monitor the error margin using a lower bound on minimum SVD loss to trigger restarts | TIMERS · MATH link prediction | relative error of MSE (%): 2.68 -> 1.14 | Improved |
| Monitor the error margin using a lower bound on minimum SVD loss to trigger restarts | TIMERS · WIKI link prediction | relative error of MSE (%): 3.44 -> 1.06 | Improved |
| Monitor the error margin using a lower bound on minimum SVD loss to trigger restarts | TIMERS · DBLP link prediction | relative error of MSE (%): 0.32 -> 0.18 | Improved |
| Monitor the error margin using a lower bound on minimum SVD loss to trigger restarts | TIMERS · INTERNET link prediction | relative error of MSE (%): 23.36 -> 11.27 | Improved |
| Use TIMERS rather than Heu-FL restart timing | Dynamic network reconstruction on FACEBOOK · FACEBOOK | maximum relative reconstruction error: 0.025 -> 0.014 | Improved |
| Use TIMERS rather than Heu-FL restart timing | Dynamic network reconstruction on MATH · MATH | maximum relative reconstruction error: 0.117 -> 0.085 | Improved |
| Use TIMERS rather than Heu-FL restart timing | Dynamic network reconstruction on WIKI · WIKI | maximum relative reconstruction error: 0.240 -> 0.139 | Improved |
| Use TIMERS rather than Heu-FL restart timing | Dynamic network reconstruction on DBLP · DBLP | maximum relative reconstruction error: 0.198 -> 0.121 | Improved |
| Use TIMERS rather than Heu-FL restart timing | Dynamic network reconstruction on INTERNET · INTERNET | maximum relative reconstruction error: 0.647 -> 0.385 | Improved |
Closest papers by meaning
- Ranking and synchronization from pairwise measurements via SVD (JMLR 2021 · similarity 0.83)
- Continuous-Time Graph Representation with Sequential Survival Process (AAAI 2024 · similarity 0.82)
- Spectral embedding for dynamic networks with stability guarantees (NeurIPS 2021 · similarity 0.82)
- EasyDGL: Encode, Train and Interpret for Continuous-Time Dynamic Graph Learning (TPAMI 2024 · similarity 0.81)
- SGD vs GD: Rank Deficiency in Linear Networks (NeurIPS 2024 · similarity 0.81)
- Long-time asymptotics of noisy SVGD outside the population limit (ICLR 2025 · similarity 0.81)
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.
