Research map / Optimization / Implicit bias and generalization
On Non-local Convergence Analysis of Deep Linear Networks
Kun Chen, Dachao Lin, Zhihua Zhang
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 |
|---|---|---|---|
| Initialization with a1(0)+b1(0)=0 (indicator zero) versus nonzero | gradient flow in deep linear networks with one-neuron layer · theory, quadratic loss | convergence to global minimizer | Mixed |
| Rank-stable set R_b(Z) initialization (s>b, u^T Z v>b) versus arbitrary balanced init | gradient flow · theory, almost-sure convergence | convergence to global minimizer | Improved |
| Gradient descent with small learning rate in saddle-init setting versus ideal gradient flow | gradient descent · synthetic deep linear net N=6, d=5 | escape from saddle | No clear effect |
Closest papers by meaning
- A Non-Local Convergence Analysis of Gradient Flow for Deep Linear Networks (TPAMI 2026 · similarity 0.98)
- A Convergence Analysis of Gradient Descent for Deep Linear Neural Networks (ICLR 2019 · similarity 0.91)
- Training Linear Neural Networks: Non-Local Convergence and Complexity Results (ICML 2020 · similarity 0.90)
- Global Convergence of Gradient Descent for Deep Linear Residual Networks (NeurIPS 2019 · similarity 0.90)
- The Loss Landscape of Deep Linear Neural Networks: a Second-order Analysis (JMLR 2024 · similarity 0.90)
- Deep Linear Network Training Dynamics from Random Initialization: Data, Width, Depth, and Hyperparameter Transfer (ICML 2025 · similarity 0.87)
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.
