Research map / Learning theory and algorithms / Sample complexity
Benefits of Additive Noise in Composing Classes with Bounded Capacity
Alireza Fathollah Pour, Hassan Ashtiani
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 |
|---|---|---|---|
| Add Gaussian noise between network layers (σ ≈ 10^-240) | Fully connected sigmoid networks · MNIST | NVAC | Improved |
| Add Gaussian noise with σ=0.2 | Fully connected sigmoid networks · MNIST | train and test accuracy | No clear effect |
| Use the proposed noisy-network covering bound instead of the comparison bounds | Fully connected neural networks · MNIST | NVAC | Improved |
Closest papers by meaning
- Generalization Bounds for Neural Networks via Approximate Description Length (NeurIPS 2019 · similarity 0.87)
- On the Role of Noise in the Sample Complexity of Learning Recurrent Neural Networks: Exponential Gaps for Long Sequences (NeurIPS 2023 · similarity 0.85)
- Near-Optimal Bounds for Learning Gaussian Halfspaces with Random Classification Noise (NeurIPS 2023 · similarity 0.84)
- Correlated Noise Provably Beats Independent Noise for Differentially Private Learning (ICLR 2024 · similarity 0.83)
- Stochastic Zeroth-Order Optimization Under Heavy-Tailed Noise (NeurIPS 2026 · similarity 0.83)
- Position: A Theory of Deep Learning Must Include Compositional Sparsity (ICML 2025 · similarity 0.83)
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.
