Regularized Learning with Networks of Features

  title={Regularized Learning with Networks of Features},
  author={Ted Sandler and John Blitzer and Partha Pratim Talukdar and Lyle H. Ungar},
For many supervised learning problems, we possess prior kno wledge about which features yield similar information about the target variab le. In predicting the topic of a document, we might know that two words are synonyms, and w hen performing image recognition, we know which pixels are adjacent. Su ch synonymous or neighboring features are near-duplicates and should be exp cted to have similar weights in an accurate model. Here we present a framework for egularized learning when one has prior… CONTINUE READING

From This Paper

Figures, tables, and topics from this paper.


Publications citing this paper.

95 Citations

Citations per Year
Semantic Scholar estimates that this publication has 95 citations based on the available data.

See our FAQ for additional information.


Publications referenced by this paper.

Similar Papers

Loading similar papers…