Research map / Optimization / Nonconvex and smooth optimization
Enhance Curvature Information by Structured Stochastic Quasi-Newton Methods
Check 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 a quasi-Newton refinement matrix to the subsampled Newton method | SSN · rcv1 and news20 | relative error | Helped |
| Use structured quasi-Newton refinement with partial Hessian information | SGD and Adam · MNIST, CURVES, and FACES autoencoders | training loss and testing loss | Helped |
| Use structured quasi-Newton refinement with partial Hessian information | KFAC · MNIST, CURVES, and FACES autoencoders | training loss and testing loss | Helped |
| Use structured quasi-Newton refinement | Adam and SGD · ConvNet on CIFAR-10 | training loss and testing accuracy | Helped |
| Use structured quasi-Newton refinement | KFAC · ResNet-18 on CIFAR-10 | training loss | Helped |
| Use structured quasi-Newton refinement | KFAC · ResNet-18 on CIFAR-10 | testing accuracy | Mixed |
| Use second-order methods rather than first-order methods | Adam and SGD · ResNet-18 on CIFAR-10 | training loss and testing accuracy | Helped |
Closest papers by meaning
- A Regularized Newton Method for Nonconvex Optimization with Global and Local Complexity Guarantees (NeurIPS 2025 · similarity 0.88)
- Skip the Hessian, Keep the Rates: Globalized Semismooth Newton with Lazy Hessian Updates (NeurIPS 2026 · similarity 0.88)
- Advancing the Lower Bounds: an Accelerated, Stochastic, Second-order Method with Optimal Adaptation to Inexactness (ICLR 2024 · similarity 0.86)
- Stochastic Newton Proximal Extragradient Method (NeurIPS 2024 · similarity 0.86)
- FOSI: Hybrid First and Second Order Optimization (ICLR 2024 · similarity 0.86)
- Single-Loop Stochastic Algorithms for Difference of Max-Structured Weakly Convex Functions (NeurIPS 2024 · 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.