Understanding the difficulty of training deep feedforward neural networks
Understanding the difficulty of training deep feedforward neural networks — presented at AISTATS 2010
Why it matters
Showed that activation functions and initialization determine training stability. The origin of Glorot initialization
Source — Primary sources first; secondary reports are labelled “Press”
Paperproceedings.mlr.presshttp://proceedings.mlr.press/v9/glorot10a.htmlAdded in Week 35, 2026 · See that week