Research map / Architectures and efficiency / RNNs and normalization
Minimum Width for Universal Approximation
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 an encoder–memorizer–decoder coding construction | ReLU networks · L^p(R^d_x, R^d_y) | minimum width for universal approximation: d_x + d_y + 1 -> max{d_x+1,d_y} | Helped |
| Use ReLU alone rather than ReLU+Step | Networks approximating continuous functions · C(K, R^d_y) | minimum width for universal approximation: max {d_x+1, d_y} -> 3 | Hurt |
| Add Step activation to ReLU | ReLU networks approximating continuous functions · C(K, R^d_y) | minimum width for universal approximation: w_min = max {d_x+1, d_y} | Helped |
| Use the coding construction with a general continuous nonpolynomial activation | General activation networks · L^p(K, R^d_y) | upper bound on minimum width: d_x+d_y+1 -> max{d_x+2,d_y+1} | Helped |
Closest papers by meaning
- Minimum Width for Universal Approximation using Squashable Activation Functions (ICML 2025 · similarity 0.93)
- Minimum width for universal approximation using ReLU networks on compact domain (ICLR 2024 · similarity 0.93)
- ReLU Network with Width $d+\mathcal{O}(1)$ Can Achieve Optimal Approximation Rate (ICML 2024 · similarity 0.92)
- Characterizing ResNet's Universal Approximation Capability (ICML 2024 · similarity 0.87)
- Minimum Width for Deep, Narrow MLP: A Diffeomorphism Approach (NeurIPS 2025 · similarity 0.85)
- On the Expressiveness of Rational ReLU Neural Networks With Bounded Depth (ICLR 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.