PubMed Health⌕ Search
Home/Archive/PubMed

PubMed · 9804675

Complexity issues in natural gradient descent method for training multilayer perceptrons.

H H Yang·S Amari
View original source ↗Export metadata ↓

Abstract

The natural gradient descent method is applied to train an n-m-1 multilayer perceptron. Based on an efficient scheme to represent the Fisher information matrix for an n-m-1 stochastic multilayer perceptron, a new algorithm is proposed to calculate the natural gradient without inverting the Fisher information matrix explicitly. When the input dimension n is much larger than the number of hidden neurons m, the time complexity of computing the natural gradient is O(n).

Explore related subjects

LearningModels, NeurologicalNeural Networks, ComputerNeuronsStochastic ProcessesJournal Article

Keep this discovery

Explore connections, maps & timelines

BibTeXRIS

H H Yang, S Amari. 1998-11-15. Complexity issues in natural gradient descent method for training multilayer perceptrons.. https://doi.org/10.1162/089976698300017007

Cite the original work for its findings. Save a collection to share your selection of sources.

KEEP EXPLORING

Related citations

PubMed2006

A statistical property of multiagent learning based on Markov decision process.

We exhibit an important property called the asymptotic equipartition property (AEP) on empirical sequences in an ergodic multiagent Markov decision process (MDP). Using the AEP which facilitates the analysis of multiagent learning, we give a statistical property of multiagent learning, such as reinforcement learning (RL), near the end of the learning process. We examine the effect of the conditions among the agents on the achievement of a cooperative policy in three different cases: blind, visible, and communicable. Also, we derive a bound on the speed with which the empirical sequence converges to the best sequence in probability, so that the multiagent learning yields the best cooperative result.

Learning↗
PubMed2006

Second order neurons and learning in Cohen-Grossberg networks.

The well known Cohen-Grossberg network is modified to include second order neural interconnections and also to have a learning component. Sufficient conditions are obtained for the existence of a globally exponentially stable equilibrium. The model provides a two-fold generalization of the Cohen-Grossberg network in the sense if one removes the learning component, then one gets a network with second order synaptic interactions; if both the learning component and the second order interactions are removed, then the model reduces to the standard Cohen-Grossberg network.

Learning↗