Stochastic gradient learning and instability: an example

Kód výsledku v IS VaVaI

<a href="https://www.isvavai.cz/riv?ss=detail&h=RIV%2F00216208%3A11640%2F16%3A00459193" target="_blank" >RIV/00216208:11640/16:00459193 - isvavai.cz</a>

Výsledek na webu

<a href="http://dx.doi.org/10.1017/S1365100514000583" target="_blank" >http://dx.doi.org/10.1017/S1365100514000583</a>

DOI - Digital Object Identifier

<a href="http://dx.doi.org/10.1017/S1365100514000583" target="_blank" >10.1017/S1365100514000583</a>

Jazyk výsledku

angličtina

Název v původním jazyce

Stochastic gradient learning and instability: an example

Popis výsledku v původním jazyce

In this paper, we investigate real-time behavior of constant-gain stochastic gradient (SG) learning, using the Phelps model of monetary policy as a testing ground. We find that whereas the self-confirming equilibrium is stable under the mean dynamics in a very large region, real-time learning diverges for all but the very smallest gain values. We employ a stochastic Lyapunov function approach to demonstrate that the SG mean dynamics is easily destabilized by the noise associated with real-time learning, because its Jacobian contains stable but very small eigenvalues. We also express caution on usage of perpetual learning algorithms with such small eigenvalues, as the real-time dynamics might diverge from the equilibrium that is stable under the mean dynamics.

Název v anglickém jazyce

Stochastic gradient learning and instability: an example

Popis výsledku anglicky

In this paper, we investigate real-time behavior of constant-gain stochastic gradient (SG) learning, using the Phelps model of monetary policy as a testing ground. We find that whereas the self-confirming equilibrium is stable under the mean dynamics in a very large region, real-time learning diverges for all but the very smallest gain values. We employ a stochastic Lyapunov function approach to demonstrate that the SG mean dynamics is easily destabilized by the noise associated with real-time learning, because its Jacobian contains stable but very small eigenvalues. We also express caution on usage of perpetual learning algorithms with such small eigenvalues, as the real-time dynamics might diverge from the equilibrium that is stable under the mean dynamics.

Druh

J_x - Nezařazeno - Článek v odborném periodiku (Jimp, Jsc a Jost)

CEP obor

AH - Ekonomie

OECD FORD obor

—

Projekt

—

Návaznosti

I - Institucionalni podpora na dlouhodoby koncepcni rozvoj vyzkumne organizace

Rok uplatnění

2016

Kód důvěrnosti údajů

S - Úplné a pravdivé údaje o projektu nepodléhají ochraně podle zvláštních právních předpisů

Název periodika

Macroeconomic Dynamics

ISSN

1365-1005

e-ISSN

—

Svazek periodika

20

Číslo periodika v rámci svazku

3

Stát vydavatele periodika

GB - Spojené království Velké Británie a Severního Irska

Počet stran výsledku

14

Strana od-do

777-790

Kód UT WoS článku

000374145600007

EID výsledku v databázi Scopus

2-s2.0-84955559951