• Brittan Farmer
  • Wednesday, November 17th. 4-5 p.m.
  • EH 3866

Sparse approximation provides the mathematical foundation for the compression of signals. One compresses a signal by finding an accurate and concise (or sparse) representation. However, the sparser the representation, the less accurate it will be. In this talk, we will define three different sparse approximation problems, each involving a different tradeoff between accuracy and sparsity. We will present algorithms for solving these problems and discuss sufficient conditions for these algorithms to recover the sparsest representation of a signal. If time permits, we will show the connection between sparse approximation and compressed sensing, in which one seeks to recover the original signal from a sparse representation.