Note on Learning Rate Schedules for Stochastic Optimization,
Abstract
We present and compare learning rate schedules for stochastic gradient descent, a general algorithm which includes LMS, on-line back-propagation and k-means clustering as special cases. We introduce search-then-converge type schedules which outperform the classical constant and running average (l/t) schedules both in speed of convergence and quality of solution.
Document Details
- Document Type
- Technical Report
- Publication Date
- Jan 01, 1992
- Accession Number
- ADP007159
Entities
People
- Christian J. Darken
- John Moody
Organizations
- Yale University