Research map / Optimization / Implicit bias and generalization
Annihilation of Spurious Minima in Two-Layer ReLU Networks
Yossi Arjevani, Michael Field
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 one neuron to the ΔS_d type I network | Two-layer ReLU network · ΔS_d type I critical points | Hessian stability: minima -> non-degenerate saddles; (d-1)-dimensional eigenspace of descent directions | Improved |
| Add one neuron to the Δ(S_{d-1} × S_1) type II network | Two-layer ReLU network · Δ(S_{d-1} × S_1) type II critical points | Hessian stability: 0.0908 -> 0.0230 | No clear effect |
| Add two neurons to the Δ(S_{d-1} × S_1) type II network | Two-layer ReLU network · Δ(S_{d-1} × S_1) type II critical points | Hessian stability: -0.1198...; at least a (d-2)-dimensional space of descent directions | Improved |
Closest papers by meaning
- Analytic Study of Families of Spurious Minima in Two-Layer ReLU Neural Networks: A Tale of Symmetry II (NeurIPS 2021 · similarity 0.96)
- Stable Minima of ReLU Neural Networks Suffer from the Curse of Dimensionality: The Neural Shattering Phenomenon (NeurIPS 2025 · similarity 0.90)
- Simplicity Bias and Optimization Threshold in Two-Layer ReLU Networks (ICML 2025 · similarity 0.90)
- Stable Minima Cannot Overfit in Univariate ReLU Networks: Generalization by Large Step Sizes (NeurIPS 2024 · similarity 0.89)
- Topological obstruction to the training of shallow ReLU neural networks (NeurIPS 2024 · similarity 0.89)
- Loss Landscape of Shallow ReLU-like Neural Networks: Stationary Points, Saddle Escape, and Network Embedding (ICLR 2025 · similarity 0.88)
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.
