Research map / Optimization / Nonconvex and smooth optimization
Accelerated Stochastic Greedy Coordinate Descent by Soft Thresholding Projection onto Simplex
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 |
|---|---|---|---|
| Add Nesterov acceleration to greedy coordinate descent (ASGCD b=n vs. non-accelerated CGD with GS-q) | greedy coordinate descent on lasso · Leukemia, Gisette, MNIST lasso (λ=1e-2, 1e-6) | log loss vs. data passes | Helped |
| Use stochastic sampling (b=1) in ASGCD instead of full-gradient greedy selection | ASGCD vs. Katyusha · Leukemia (λ=1e-2, 1e-6) and Gisette (λ=1e-2) | log loss vs. data passes | Helped |
| Use ASGCD (b=1) instead of Katyusha | ASGCD vs. Katyusha · Gisette (λ=1e-6) and MNIST (λ=1e-2, 1e-6) | log loss vs. data passes | Hurt |
| Use the l1-norm square approximation greedy rule (ASGCD b=n) instead of l2 quadratic proximal AFG | ASGCD (b=n) vs. AFG · Leukemia (λ=1e-2, 1e-6) and Gisette (λ=1e-2) | log loss vs. data passes | Helped |
| Use the l1-norm square approximation greedy rule (ASGCD b=n) instead of l2 quadratic proximal AFG | ASGCD (b=n) vs. AFG · Gisette (λ=1e-6) and MNIST (λ=1e-2, 1e-6) | log loss vs. data passes | Hurt |
| Sort-free SOTOPO filtering (Lemma 5) to reduce greedy selection cost | SOTOPO projection · theoretical complexity | cost | Helped |
Closest papers by meaning
- Accelerating Greedy Coordinate Descent Methods (ICML 2018 · similarity 0.91)
- Approximate Steepest Coordinate Descent (ICML 2017 · similarity 0.88)
- Even Faster Accelerated Coordinate Descent Using Non-Uniform Sampling (ICML 2016 · similarity 0.88)
- Optimization over Sparse Support-Preserving Sets: Two-Step Projection with Global Optimality Guarantees (ICML 2025 · similarity 0.87)
- High-probability complexity bounds for stochastic non-convex minimax optimization (NeurIPS 2024 · similarity 0.86)
- Det-CGD: Compressed Gradient Descent with Matrix Stepsizes for Non-Convex Optimization (ICLR 2024 · similarity 0.86)
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.