Research map / Optimization / Optimizers
Better SGD using Second-order Momentum
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 Hessian-corrected momentum to SGD | SGD, ResNet20 · CIFAR-10 | test accuracy: 92.14 ± 0.15 -> 92.46 ± 0.07 | Helped |
| Add Hessian-corrected momentum to SGD | SGD, ResNet32 · CIFAR-10 | test accuracy: 93.08 ± 0.12 -> 93.19 ± 0.08 | Helped |
| Add Hessian-corrected momentum to SGD | SGD, ResNet18 · ImageNet | top-1 accuracy: 70.36 -> 70.58 | Helped |
| Add Hessian-corrected momentum to SGD | SGD · IWSLT'14 German-English | BLEU: 29.75 -> 33.73 | Helped |
| Use adaptive SGDHess instead of the non-adaptive SGDHess variant | SGDHess · IWSLT'14 German-English | BLEU: 33.73; 35.53 | Mixed |
| Add Hessian-vector-product momentum correction rather than use first-order SGD | SGD, image classification · Image classification | runtime: 1.7 times slower | Hurt |
| Add Hessian-vector-product momentum correction rather than use first-order SGD/Adam | SGD/Adam, neural machine translation · Neural machine translation | runtime: 1.3 times slower | Hurt |
Closest papers by meaning
- An Improved Analysis of Stochastic Gradient Descent with Momentum (NeurIPS 2020 · similarity 0.92)
- Gradient correlation is a key ingredient to accelerate SGD with momentum (ICLR 2025 · similarity 0.91)
- The Marginal Value of Momentum for Small Learning Rate SGD (ICLR 2024 · similarity 0.91)
- Random Scaling and Momentum for Non-smooth Non-convex Optimization (ICML 2024 · similarity 0.90)
- Perfect Parallelization in Mini-Batch SGD with Classical Momentum Acceleration (NeurIPS 2026 · similarity 0.89)
- Non-asymptotic Analysis of Biased Adaptive Stochastic Approximation (NeurIPS 2024 · similarity 0.89)
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.