<p>In this paper our focus is on analysis and design of linear and polyhedral signal recoveries robust with respect to the deterministic uncertainty in the observation matrix. This can be seen as a “deterministic counterpart” of the work [<CitationRef CitationID="CR1">1</CitationRef>] where the case of random uncertainty was studied. We investigate the performance of estimates robust w.r.t. deterministic norm-bounded matrix uncertainty, derive efficiently computable bounds for the estimation risk and discuss the construction of “presumably good” estimates.</p>

错误:搜索内容不能为空,请输入英文关键词
错误:关键词超出字数限制,请精简
高级检索

Robust Signal Recovery Under Uncertain-but-Bounded Perturbations in Observation Matrix

  • Yannis Bekri,
  • Anatoli Juditsky,
  • Arkadi Nemirovski

摘要

In this paper our focus is on analysis and design of linear and polyhedral signal recoveries robust with respect to the deterministic uncertainty in the observation matrix. This can be seen as a “deterministic counterpart” of the work [1] where the case of random uncertainty was studied. We investigate the performance of estimates robust w.r.t. deterministic norm-bounded matrix uncertainty, derive efficiently computable bounds for the estimation risk and discuss the construction of “presumably good” estimates.