Research map / Optimization / Optimizers
SGLB: Stochastic Gradient Langevin Boosting
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 SGLB instead of GB while optimizing the smoothed 0-1 loss | Gradient boosting on synthetic data · Synthetic data | 0-1 loss: 0.475 -> 0.470 | Helped |
| Use SGLB instead of SGB while optimizing the smoothed 0-1 loss | Gradient boosting on synthetic data · Synthetic data | 0-1 loss: 0.474 -> 0.470 | Helped |
| Optimize smoothed 0-1 loss rather than logistic loss | Gradient boosting on synthetic data · Synthetic data | 0-1 loss: 0.482 -> 0.470 | Helped |
| Use SGLB instead of SGB | Gradient boosting · Appetency (0-1 loss) | 0-1 loss: 1.8 -> 1.8 | No clear effect |
| Use SGLB instead of SGB | Gradient boosting · Churn (0-1 loss) | 0-1 loss: 7.1 -> 7.2 | Hurt |
| Use SGLB instead of SGB | Gradient boosting · Upselling (0-1 loss) | 0-1 loss: 4.8 -> 4.7 | Helped |
| Use SGLB instead of SGB | Gradient boosting · Adult (0-1 loss) | 0-1 loss: 13.2 -> 12.8 | Helped |
| Use SGLB instead of SGB | Gradient boosting · Amazon (0-1 loss) | 0-1 loss: 5.2 -> 4.8 | Helped |
| Use SGLB instead of SGB | Gradient boosting · Click (0-1 loss) | 0-1 loss: 16 -> 15.9 | Helped |
| Use SGLB instead of SGB | Gradient boosting · Epsilon (0-1 loss) | 0-1 loss: 11.7 -> 11.7 | No clear effect |
| Use SGLB instead of SGB | Gradient boosting · Higgs (0-1 loss) | 0-1 loss: 25.2 -> 24.8 | Helped |
| Use SGLB instead of SGB | Gradient boosting · Internet (0-1 loss) | 0-1 loss: 10.1 -> 9.8 | Helped |
Closest papers by meaning
- The promises and pitfalls of Stochastic Gradient Langevin Dynamics (NeurIPS 2018 · similarity 0.89)
- A Contour Stochastic Gradient Langevin Dynamics Algorithm for Simulations of Multi-modal Distributions (NeurIPS 2020 · similarity 0.88)
- Stochastic Approximate Gradient Descent via the Langevin Algorithm (AAAI 2020 · similarity 0.88)
- Minimal Variance Sampling in Stochastic Gradient Boosting (NeurIPS 2019 · similarity 0.87)
- Improved Convergence Rate of Stochastic Gradient Langevin Dynamics with Variance Reduction and its Application to Optimization (NeurIPS 2022 · similarity 0.87)
- Investigating the Role of Weight Decay in Enhancing Nonconvex SGD (CVPR 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.