Homotopy algorithm using dichotomous coordinate descent iterations for sparse recovery

Autor(es)

Zakharov, Yuriy ; Nascimento, Vitor H. ; Matthews, MB

• Tipo de documento: Artigo Científico
• Fonte: 2012 CONFERENCE RECORD OF THE FORTY SIXTH ASILOMAR CONFERENCE ON SIGNALS, SYSTEMS AND COMPUTERS (ASILOMAR); v. N/A, p. 5-pg., 2012-01-01.

Resumo

We propose a computationally efficient technique for recovery of sparse signals, which is based on dichotomous coordinate descent iterations, homotopy, and reweighting. We derive optimal rules for adding and removing elements to/from the support, guaranteeing that the cost function is minimized. Numerical examples show that the proposed technique achieves a mean-squared error smaller than that of the YALLI algorithm and complexity comparable to that of the matched pursuit algorithm. (AU)

• Processo FAPESP: 12/50565-1 - Adaptive compressive sensing-aware techniques: desinf algorithms and applications
• Beneficiário: Vitor Heloiz Nascimento
• Modalidade de apoio: Auxílio à Pesquisa - Regular
• Processo FAPESP: 11/06994-2 - Algoritmos de baixo custo computacional para processamento de sinais acústicos
• Beneficiário: Vitor Heloiz Nascimento
• Modalidade de apoio: Auxílio à Pesquisa - Regular