Efficient Numerical Algorithm for Large-Scale Damped Natural Gradient Descent

Han Wang; Hao Xie; Yixiao Chen

REVIEW

Not yet reviewed by Pith; the record is open.

Re-run · record.json Download PDF Read on arXiv ↗

This paper has not been read by Pith yet. Machine review is queued; the pith claim, tier, and objections will appear here once it completes.

SPECIMEN: schema-true, not a live event

T0 review · schema-true

One-sentence machine reading of the paper's core claim.

pith:XXXXXXXX · record.json · timestamp

arxiv 2310.17556 v1 pith:45WMXFJE submitted 2023-10-26 cs.LG cs.NAmath.NAphysics.comp-ph

Efficient Numerical Algorithm for Large-Scale Damped Natural Gradient Descent

Yixiao Chen , Hao Xie , Han Wang This is my paper

classification cs.LG cs.NAmath.NAphysics.comp-ph

keywords algorithmdampeddescentgradientlarge-scalenaturalnumbersignificantly

verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved

0 comments

read the original abstract

We propose a new algorithm for efficiently solving the damped Fisher matrix in large-scale scenarios where the number of parameters significantly exceeds the number of available samples. This problem is fundamental for natural gradient descent and stochastic reconfiguration. Our algorithm is based on Cholesky decomposition and is generally applicable. Benchmark results show that the algorithm is significantly faster than existing methods.

Efficient Numerical Algorithm for Large-Scale Damped Natural Gradient Descent

discussion (0)