Research map / Optimization / Optimizers
Nonparametric Neural Networks
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 the nonparametric network instead of a same-size parametric network | Four-hidden-layer networks · poker test set; size 23-24-15-4 | classification error: 0.20% -> 0.62% | Hurt |
| Use the nonparametric network instead of a same-size parametric network | Four-hidden-layer networks · poker test set; size 94-135-105-35 | classification error: 0.003% -> 0.022% | Helped |
| Use the nonparametric network instead of a same-size parametric network | Four-hidden-layer networks · poker test set; size 210-251-224-104 | classification error: 0.003% -> 0.001% | Helped |
| Use the nonparametric network instead of a same-size parametric network | Four-hidden-layer networks · poker test set; size 299-258-259-129 | classification error: 0.002% -> 0% | Helped |
| Adapt network size during training rather than train a fixed-size network from scratch | Nonparametric neural networks · MNIST | test classification error | Helped |
| Adapt network size during training rather than train a fixed-size network from scratch | Nonparametric neural networks · rectangles images | test classification error | Mixed |
| Adapt network size during training rather than train a fixed-size network from scratch | Nonparametric neural networks · convex | test classification error | Hurt |
| Add momentum using AdaRad-M (ablation) | AdaRad · Experiments across datasets | performance | No clear effect |
| Use Adam instead of RMSprop (ablation) | Parametric networks · Experiments across datasets | performance | No clear effect |
Closest papers by meaning
- Deep Learning meets Nonparametric Regression: Are Weight-Decayed DNNs Locally Adaptive? (ICLR 2023 · similarity 0.86)
- Porcupine Neural Networks: Approximating Neural Network Landscapes (NeurIPS 2018 · similarity 0.86)
- On the Hardness of Training Deep Neural Networks Discretely (AAAI 2025 · similarity 0.85)
- The Power of Random Features and the Limits of Distribution-Free Gradient Descent (ICML 2025 · similarity 0.84)
- Random Sparse Lifts: Construction, Analysis and Convergence of finite sparse networks (ICLR 2024 · similarity 0.84)
- Hybrid Decentralized Optimization: Leveraging Both First- and Zeroth-Order Optimizers for Faster Convergence (AAAI 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.