Relational Lasso - An Improved Method Using the Relations Among Features -

Relational lasso is a method that incorporates feature relations within machine learning. By using automatically obtained noisy relations among features, relational lasso learns an additional penalty parameter per feature, which is then incorporated in terms of a regularizer within the target optimization function. Relational lasso has been tested on three different tasks: text categorization, polarity estimation, and parsing, where it was compared with conventional lasso and adaptive lasso (Zou, 2006) when using a multi-class logistic regression optimization method. Relational lasso outperformed these other lasso methods in the tests.