Research map / Optimization / Nonconvex and smooth optimization
RECAPP: Crafting a More Efficient Catalyst for Convex Optimization
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 RECAPP with SVRG instead of Catalyst-SVRG | Finite-sum convex minimization · Finite-sum convex minimization | gradient query complexity: O\prn*{ (n + \sqrt{nLR^2/\epsilon})\log \frac {LR^2}{\epsilon}} -> O(n\log\log n+\sqrt{nLR^2 /\epsilon}) | Helped |
| Use RECAPP-Mirror-Prox instead of the Catalyst-based method | Max-structured minimization · Max-structured minimization | gradient query complexity: O{\prn*{\frac{L R}{\sqrt{\mu \epsilon}} \log \frac{L R}{\sqrt{\mu \epsilon}}}} -> O\prn*{\frac{LR}{\sqrt{\mu\epsilon}}} | Helped |
| Use RECAPP instead of Catalyst C1* | Logistic regression · Covertype | runtime | Helped |
| Use RECAPP instead of Catalyst C1* | Logistic regression · A9a | runtime | No clear effect |
| Use RECAPP instead of Catalyst C1* | Logistic regression · Real-sim | runtime | Hurt |
| Enable MLMC debiasing with p∈{0.1,0.25} instead of p=0 (ablation) | RECAPP · Logistic regression datasets | runtime | Helped |
| Remove MLMC debiasing (p=0) (ablation) | RECAPP · Logistic regression datasets | runtime | Hurt |
Closest papers by meaning
- A Catalyst Framework for Minimax Optimization (NeurIPS 2020 · similarity 0.89)
- Minibatch Stochastic Approximate Proximal Point Methods (NeurIPS 2020 · similarity 0.89)
- Accelerated Quasi-Newton Proximal Extragradient: Faster Rate for Smooth Convex Optimization (NeurIPS 2023 · similarity 0.89)
- Adaptive Proximal Gradient Method for Convex Optimization (NeurIPS 2024 · similarity 0.88)
- Learning Proximal Operators to Discover Multiple Optima (ICLR 2023 · similarity 0.88)
- Epigraph projections for fast general convex programming (ICML 2016 · similarity 0.87)
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.