paper lists and information on mean-field theory of deep learning
-
Updated
Mar 25, 2019
#
dynamical-isometry
Here are 3 public repositories matching this topic...
Mathematical consequences of orthogonal weights initialization and regularization in deep learning. Experiments with gain-adjusted orthogonal regularizer on RNNs with SeqMNIST dataset.
keras mathematics rnn mathematics-machine-learning neuralnetworks orthogonal initialization dynamical-isometry
-
Updated
Sep 21, 2019 - Jupyter Notebook
Make your deep neural network Dynamically Isometric. Farewell to vanishing and exploding gradients
-
Updated
Jul 8, 2022 - Python
Add this topic to your repo
To associate your repository with the dynamical-isometry topic, visit your repo's landing page and select "manage topics."