Recursive l(sub 1, infinity) Group lasso

Abstract

We introduce a recursive adaptive group lasso algorithm for real-time penalized least squares prediction that produces a time sequence of optimal sparse predictor coefficient vectors. At each time index the proposed algorithm computes an exact update of the optimal L(sub 1, infinity)-penalized recursive least squares (RLS) predictor. Each update minimizes a convex but nondifferentiable function optimization problem. We develop an online homotopy method to reduce the computational complexity. Numerical simulations demonstrate that the proposed algorithm outperforms the l(sub 1) regularized RLS algorithm for a group sparse system identification problem and has lower implementation complexity than direct group lasso solvers.

Open PDF

Document Details

Document Type
Technical Report
Publication Date
Aug 01, 2012
Accession Number
ADA588199

Entities

People

  • Alfred O. Hero III
  • Yilun Chen

Organizations

  • University of Michigan

Tags

Communities of Interest

  • Materials and Manufacturing Processes

DTIC Thesaurus Topics

  • Adaptive Filters
  • Algorithms
  • Coefficients
  • Compressed Sensing
  • Computational Complexity
  • Computations
  • Computer Science
  • Electrical Engineering
  • Filters
  • Identification
  • Iterations
  • Microarchitecture
  • Notation
  • Optimization
  • Signal Processing
  • Simulations
  • Steady State

Fields of Study

  • Computer science

Readers

  • Approximation Theory.
  • Neural Network Machine Learning.
  • Operations Research