# A Novel Structured Natural Gradient Descent for Deep Learning

arXiv: https://arxiv.org/abs/2109.10100 · HTML: https://mlsanity.com/paper/2109.10100/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Builds on 9 of its 21 references that are in the map.
- First posted on arXiv 21 Sep 2021.

## Builds on (most cited first, 9 of 9)

- [Batch Normalization: Accelerating Deep Network Training by Reducing Internal Covariate Shift](https://mlsanity.com/paper/1502.03167/) (2015)
- [Spectral Normalization for Generative Adversarial Networks](https://mlsanity.com/paper/1802.05957/) (2018)
- [Group Normalization](https://mlsanity.com/paper/1803.08494/) (2018)
- [Semantic Image Synthesis with Spatially-Adaptive Normalization](https://mlsanity.com/paper/1903.07291/) (2019)
- [Instance Normalization: The Missing Ingredient for Fast Stylization](https://mlsanity.com/paper/1607.08022/) (2016)
- [Weight Normalization: A Simple Reparameterization to Accelerate Training of Deep Neural Networks](https://mlsanity.com/paper/1602.07868/) (2016)
- [Fast Approximate Natural Gradient Descent in a Kronecker-factored Eigenbasis](https://mlsanity.com/paper/1806.03884/) (2018)
- [Fast Convergence of Natural Gradient Descent for Overparameterized Neural Networks](https://mlsanity.com/paper/1905.10961/) (2019)
- [Improved Bilinear Pooling with CNNs](https://mlsanity.com/paper/1707.06772/) (2017)

## Abstract (arXiv)

Natural gradient descent (NGD) provided deep insights and powerful tools to deep neural networks. However the computation of Fisher information matrix becomes more and more difficult as the network structure turns large and complex. This paper proposes a new optimization method whose main idea is to accurately replace the natural gradient optimization by reconstructing the network. More specifically, we reconstruct the structure of the deep neural network, and optimize the new network using traditional gradient descent (GD). The reconstructed network achieves the effect of the optimization way with natural gradient descent. Experimental results show that our optimization method can accelerate the convergence of deep network models and achieve better performance than GD while sharing its computational simplicity.