Evans, George W.
[VerfasserIn]
;
Williams, Noah
[Sonstige Person, Familie und Körperschaft];
Honkapohja, Seppo
[Sonstige Person, Familie und Körperschaft]National Bureau of Economic Research
Erschienen:
Cambridge, Mass: National Bureau of Economic Research, October 2005
Erschienen in:NBER technical working paper series ; no. t0317
Umfang:
1 Online-Ressource
Sprache:
Englisch
DOI:
10.3386/t0317
Identifikator:
Reproduktionsnotiz:
Hardcopy version available to institutional subscribers
Entstehung:
Anmerkungen:
Mode of access: World Wide Web
System requirements: Adobe [Acrobat] Reader required for PDF files
Beschreibung:
We study the properties of generalized stochastic gradient (GSG) learning in forward-looking models. We examine how the conditions for stability of standard stochastic gradient (SG) learning both differ from and are related to E-stability, which governs stability under least squares learning. SG algorithms are sensitive to units of measurement and we show that there is a transformation of variables for which E-stability governs SG stability. GSG algorithms with constant gain have a deeper justification in terms of parameter drift, robustness and risk sensitivity