# A Common Framework for Natural Gradient and Taylor based Optimisation using Manifold Theory

arXiv: https://arxiv.org/abs/1803.09791 · HTML: https://mlsanity.com/paper/1803.09791/

## Key facts

- Cited by 1 papers in the map (data through 14 Oct 2026).
- First posted on arXiv 26 Mar 2018.

## Built on by (most cited first, 1 of 1)

- [A Distributed Optimisation Framework Combining Natural Gradient with Hessian-Free for Discriminative Sequence Training](https://mlsanity.com/paper/2103.07554/) (2021)

## Abstract (arXiv)

This technical report constructs a theoretical framework to relate standard Taylor approximation based optimisation methods with Natural Gradient (NG), a method which is Fisher efficient with probabilistic models. Such a framework will be shown to also provide mathematical justification to combine higher order methods with the method of NG.