# TENGraD: Time-Efficient Natural Gradient Descent with Exact Fisher-Block Inversion

Source: [TENGraD: Time-Efficient Natural Gradient Descent with Exact Fisher-Block Inversion](<https://www.sota2.com/research/paper/68fa6bf51e209131cea3136c>)

Collection index: [llms.txt](<https://www.sota2.com/research/llms.txt>)

Data updated: 2025-10-27T06:13:32.751Z

Authors: Saeed Soori, Bugra Can, Baourun Mu, Mert G\\"urb\\"uzbalaban, Maryam Mehri Dehnavi

Published: 2021-06-07T20:16:15.000Z

## Abstract

This work proposes a time-efficient Natural Gradient Descent method, called TENGraD, with linear convergence guarantees. Computing the inverse of the neural network's Fisher information matrix is expensive in NGD because the Fisher matrix is large. Approximate NGD methods such as KFAC attempt to improve NGD's running time and practical application by reducing the Fisher matrix inversion cost with approximation. However, the approximations do not reduce the overall time significantly and lead to less accurate parameter updates and loss of curvature information. TENGraD improves the time efficiency of NGD by computing Fisher block inverses with a computationally efficient covariance factorization and reuse method. It computes the inverse of each block exactly using the Woodbury matrix identity to preserve curvature information while admitting (linear) fast convergence rates. Our experiments on image classification tasks for state-of-the-art deep neural architecture on CIFAR-10, CIFAR-100, and Fashion-MNIST show that TENGraD significantly outperforms state-of-the-art NGD methods and often stochastic gradient descent in wall-clock time.

## Recorded benchmark results

A research result applies to its stated dataset, split, metric and evaluation configuration. Compare like-for-like settings, preserve the metric's unit and direction, and distinguish paper-reported results from independent reproduction. The best result recorded here is not a claim of exhaustive, current global coverage.

## Original sources and implementation

[Original paper](<https://arxiv.org/abs/2106.03947>)

[PDF](<https://arxiv.org/pdf/2106.03947.pdf>)
