Research map / Optimization / Nonconvex and smooth optimization
Mollifying 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 mollification with annealed noise and stochastic identity paths | 110-layer residual convolutional network · CIFAR10 test | accuracy: 91.78 -> 92.45 | Helped |
| Use mollification with annealed noise and stochastic identity paths | 110-layer residual convolutional network · CIFAR10 test | accuracy: 93.25 -> 92.45 | Hurt |
| Use mollified training instead of vanilla residual connections | 6-layer sigmoid MLP · Pentomino 80k, 100 epochs | accuracy: 69.5% -> 75.15% | Helped |
| Use mollified training instead of a standard 2-layer LSTM | 2-layer LSTM · PTB word-level language modeling | PPL: 128.4 -> 123.6 | Helped |
| Use mollified training rather than the standard deep MLP training setup | 6-layer sigmoid MLP · 40-dimensional parity | convergence speed | Helped |
Closest papers by meaning
- On the Hardness of Training Deep Neural Networks Discretely (AAAI 2025 · similarity 0.86)
- On the Quality of the Initial Basin in Overspecified Neural Networks (ICML 2016 · similarity 0.86)
- No Wrong Turns: The Simple Geometry Of Neural Networks Optimization Paths (ICML 2024 · similarity 0.85)
- Loss Landscape Characterization of Neural Networks without Over-Parametrization (NeurIPS 2024 · similarity 0.85)
- Improving Generalization of Deep Neural Networks by Optimum Shifting (AAAI 2025 · similarity 0.85)
- On the Parameterization of Second-Order Optimization Effective towards the Infinite Width (ICLR 2024 · 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.