Regularization

From RecSysWiki
Revision as of 12:28, 6 June 2011 by Zeno Gantner (talk | contribs) (Created page with "'''Regularization''' methods penalize high values for model parameters in order to avoid overfitting effects. For example the optimization (minimization) problem for a [[...")
(diff) ← Older revision | Latest revision (diff) | Newer revision → (diff)
Jump to navigation Jump to search

Regularization methods penalize high values for model parameters in order to avoid overfitting effects. For example the optimization (minimization) problem for a matrix factorization model used in a recommender system typically contains a penalty term for the elements of the lower-rank matrices used to approximate the user-item matrix, e.g. via the Frobenius norm.

External links