phenomenon · first seen 30 Oct 2026
symmetry breaking in neural network training
Neurons starting from identical or small initializations differentiate and specialize to distinct features during gradient training.
Mentions per week
PapersCodeDiscussion
Earliest sources
- Gradient Descent on Two ReLU Neurons: Global Landscape and Bifurcation DynamicsPapers · 30 Oct 2026 · Binghua Li, Mengzhe Li, Denny Wu, Tianhao Wang
Latest sources
- Gradient Descent on Two ReLU Neurons: Global Landscape and Bifurcation DynamicsPapers · 30 Oct 2026 · Binghua Li, Mengzhe Li, Denny Wu, Tianhao Wang