Research map / Optimization / Optimizers
Training Deep Models Faster with Robust, Approximate Importance Sampling
Tyler B Johnson, Carlos Guestrin
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 RAIS importance sampling with adaptive learning rate instead of uniform SGD | SGD · SVHN | validation error: 0.0226 -> 0.0201 | Improved |
| use RAIS importance sampling with adaptive learning rate instead of uniform SGD | SGD · rot-MNIST | validation error: 0.0512 -> 0.0476 | Improved |
| use RAIS importance sampling with adaptive learning rate instead of uniform SGD | SGD · CIFAR-10 | validation error: 0.0607 -> 0.0590 | Improved |
| use RAIS importance sampling with adaptive learning rate instead of uniform SGD | SGD · CIFAR-100 | validation error: 0.236 -> 0.236 | No clear effect |
| use RAIS importance sampling with adaptive learning rate instead of uniform SGD | SGD · CIFAR-100 | training speed: approximately 20% speed-up | Improved |
| use RAIS importance sampling with adaptive learning rate instead of uniform SGD | SGD · SVHN | training speed: more than 2x | Improved |
| use exact oracle importance sampling instead of RAIS | RAIS-SGD · MNIST LeNet | training objective | Improved |
Closest papers by meaning
- Not All Samples Are Created Equal: Deep Learning with Importance Sampling (ICML 2018 · similarity 0.90)
- LITE: A Lightweight Lazy Sampler for Efficient SGD (NeurIPS 2026 · similarity 0.89)
- Safe Adaptive Importance Sampling (NeurIPS 2017 · similarity 0.87)
- Reparameterized Importance Sampling for Robust Variational Bayesian Neural Networks (ICML 2024 · similarity 0.85)
- Understanding the Training Speedup from Sampling with Approximate Losses (ICML 2024 · similarity 0.84)
- LEAPS: A discrete neural sampler via locally equivariant networks (ICML 2025 · similarity 0.84)
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.
