Methods › General › Optimization › Natural Gradient Descent

Natural Gradient Descent

introduced 1998 68 papers tagged archive 2025-07-28

archive 2025-07-28 Description, source and code snippet are the archive's method entry.

Natural Gradient Descent is an approximate second-order optimisation method. It has an interpretation as optimizing over a Riemannian manifold using an intrinsic distance metric, which implies the updates are invariant to transformations such as whitening. By using the positive semi-definite (PSD) Gauss-Newton matrix to approximate the (possibly negative definite) Hessian, NGD can often work better than exact second-order methods.

Given the gradient of z, g = (δf(z))/δz, NGD computes the update as:

Δz = αF⁻¹g

where the Fisher information matrix F is defined as:

F = 𝔼_(p(t|z))[∇lnp(t|z)∇lnp(t|z)ᵀ]

The log-likelihood function lnp(t|z) typically corresponds to commonly used error functions such as the cross entropy loss.

Source: LOGAN

Image: Fast Convergence of Natural Gradient Descent for Overparameterized Neural Networks

In Syntology name not resolved (unknown_name). The lookup matched no method, paper or venue by this name; that asserts nothing about whether code exists.

Near matches the lookup lists, each a different name: natural Riemannian gradient descent; Natural Hypergradient Descent (NHGD); GAUSS-NEWTON NATURAL GRADIENT DESCENT; natural gradient descent with momentum; Tractable Structured Natural-Gradient Descent; Zeroth-Order Natural Gradient Descent (ZO-NGD); Energy Manifold Natural Gradient Descent (EMNGD); Natural Gradient Descent for Online Continual Learning.

Syntology's have check for this name (free, it answers coverage only), read 2026-10-05; ask it again live. A level is the served verification level of that one answer; nothing here is summed across methods.

Papers archive 2025-07-28

archive papers tagged: 68 · with a code link: 21 · where Syntology ran a sample: 6 (5 with a run with no instrument failure, 1 where every run was a failure of Syntology's instrument) Syntology

Show: all tagged papersonly where code ran (6 of 68 tagged: 5 with a run with no instrument failure, 1 where every run was a failure of Syntology's instrument)

Page 1 of 1: papers 1 to 68 of 68, newest first by the archive's date (ties by slug), in archive order.

Repository counts are the archive's code-links table. A Syntology line states what Syntology ran from that paper's harvested code, as “N ran (of which C constructed an object rather than computing a result; K with no instrument failure: H honoured, V violated, P with no contract checked; I where Syntology's instrument failed) · U unverified”; the instrument figure counts failures of Syntology's instrument, not of the code. It is per sample and not a correctness claim. When the archive marks a repository official for the paper, the line starts with that repository's state (the archive's flag, not a verdict on who wrote the code; “community repositories only” when every sample that ran came from a community repository, “official: no sample here; runs from other or unrecorded repositories” when some came from a repository the paper names or has in its text, or from none recorded); hover it for the repositories the samples that ran came from.

Tasks archive 2025-07-28

20 shown of 51 tasks the archive attaches to papers tagged with this method, by distinct papers. A task without a page in the catalog is plain text.

TaskPapers
Variational Inference8
Image Classification7
image-classification7
Second-order methods5
Computational Efficiency4
Deep Learning3
Quantum Machine Learning3
Stochastic Optimization3
Variational Monte Carlo3
BIG-bench Machine Learning2
Bayesian Inference2
Bias Detection2
Clustering2
Federated Learning2
Image Reconstruction2
Language Modeling2
Language Modelling2
regression2
3D Reconstruction1
Adversarial Attack1

Usage over time archive 2025-07-28

Papers per year tagged with Natural Gradient Descent: 2019 to 2025, peak 20 20 0 2019: 5 papers 2019 2020: 9 papers 2020 2021: 12 papers 2021 2022: 9 papers 2022 2023: 8 papers 2023 2024: 20 papers 2024 2025: 5 papers 2025
Papers per year the archive tags with this method, by the paper's archive date (68 dated). Bars are counts, not a trend claim.

Components: the archive holds no method-to-method composition, so PwC's Components table cannot be rebuilt; the Papers list carries no Results column for the same reason (the archive does not join its leaderboard rows to method tags).

Categories archive 2025-07-28

Optimization

Report a problem or propose a change · a person checks every report against the paper or source before anything changes; decisions are listed on /corrections