Research map / Optimization / Implicit bias and generalization
Spurious Valleys and Clustering Behavior of 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 |
|---|---|---|---|
| Hidden layer sizes set to at least the k-intrinsic dimension (h_k >= dim_k) for no-bias networks | polynomial NN without bias (theory) · theoretical sufficient condition (Theorem 3.7) | existence of spurious valleys in loss landscape | No clear effect |
| Hidden layer sizes set to at least intrinsic dimension minus one (h_k >= dim_k - 1) for networks with bias | polynomial NN with bias (theory) · theoretical sufficient condition (Theorem 3.10) | existence of spurious valleys in loss landscape | No clear effect |
| Translating training inputs toward infinity (delta to +inf) for a representable target in a 1-3-1 square-activation network | cluster behavior of cut loss landscape error values · target 3x^2-13, 1-3-1 2-NN (Example 4.1) | middle cluster error values limit | Mixed |
Closest papers by meaning
- Pure and Spurious Critical Points: a Geometric Study of Linear Networks (ICLR 2020 · similarity 0.91)
- Flat Channels to Infinity in Neural Loss Landscapes (NeurIPS 2025 · similarity 0.87)
- SAD Neural Networks: Divergent Gradient Flows and Asymptotic Optimality via o-minimal Structures (NeurIPS 2025 · similarity 0.87)
- Loss Landscape Characterization of Neural Networks without Over-Parametrization (NeurIPS 2024 · similarity 0.87)
- Exploring The Loss Landscape Of Regularized Neural Networks Via Convex Duality (ICLR 2025 · similarity 0.86)
- No Wrong Turns: The Simple Geometry Of Neural Networks Optimization Paths (ICML 2024 · similarity 0.86)
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.