Research map / Optimization / Distributed and compressed
GIANT: Globally Improved Approximate Newton Method for Distributed Optimization
Shusen Wang, Fred Roosta, Peng Xu, Michael W. Mahoney
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (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 GIANT instead of AGD, DANE, or L-BFGS | Distributed logistic regression · MNIST8M, Epsilon, and Covtype | training objective and test classification error | Improved |
| Increase the node count from 75 to 375 | GIANT · Covtype | per-iteration time increase: 19% | Hurt |
| Increase the node count from 75 to 375 | AGD · Covtype | per-iteration time increase: 367% | Hurt |
| Increase the node count from 75 to 375 | L-BFGS · Covtype | per-iteration time increase: 112% | Hurt |
| Scale the dataset while increasing compute nodes to keep local computation per iteration the same | GIANT and DANE · Augmented datasets (n, 5n, and 25n) | objective convergence | No clear effect |
| Tune the fixed line-search parameters | GIANT, L-BFGS, and DANE · Logistic regression experiments | convergence | No clear effect |
| Allow an inexact local linear-system solution with error on the order of the sampling error | GIANT · Smooth strongly convex objectives | convergence rate | No clear effect |
| Use Nesterov acceleration in DANE to form AIDE | DANE · Off-line small-scale simulations | convergence | Hurt |
Closest papers by meaning
- DINO: Distributed Newton-Type Optimization Method (ICML 2020 · similarity 0.90)
- Newton Method over Networks is Fast up to the Statistical Precision (ICML 2021 · similarity 0.88)
- Accelerated Methods with Compressed Communications for Distributed Optimization Problems Under Data Similarity (AAAI 2025 · similarity 0.86)
- Do We Need Asynchronous SGD? On the Near-Optimality of Synchronous Methods (NeurIPS 2026 · similarity 0.86)
- Communication Efficient Distributed Newton Method over Unreliable Networks (AAAI 2024 · similarity 0.86)
- Towards Faster Decentralized Stochastic Optimization with Communication Compression (ICLR 2025 · 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.
