Title :
Variable selection by RIVAL
Author :
Er-Wei Bai ; Kang Li ; Kump, Paul
Author_Institution :
Dept. of Electr. & Comput. Eng., Univ. of Iowa, Iowa City, IA, USA
Abstract :
The paper considers variable selection problem and proposes an algorithm called the RIVAL (Removing Irrelevant Variables Amidst Lasso Iterations). For a given and fixed length of data points, the algorithm recursively updates the weights so that the ability of the algorithm in detecting zero coefficients is substantially improved. Theoretical convergence is established supported by numerical simulation results.
Keywords :
iterative methods; recursive estimation; RIVAL; data point fixed length; removing irrelevant variables amidst Lasso iterations; variable selection; zero coefficients detection; Adaptation models; Convergence; Educational institutions; Indexes; Input variables; Numerical simulation; Vectors;
Conference_Titel :
Control and Automation (MED), 2014 22nd Mediterranean Conference of
Conference_Location :
Palermo
Print_ISBN :
978-1-4799-5900-6
DOI :
10.1109/MED.2014.6961490
