Efficient Numerical Algorithm for Large-Scale Damped Natural Gradient Descent

Autor:	Chen, Yixiao, Xie, Hao, Wang, Han
Rok vydání:	2023
Předmět:	Computer Science - Machine Learning Mathematics - Numerical Analysis Physics - Computational Physics
Druh dokumentu:	Working Paper
Popis:	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.
Databáze:	arXiv
Externí odkaz:	http://arxiv.org/abs/2310.17556 Zobrazit plný text záznamu View this record from Arxiv