An Inverse problem where the measurements are linearly related to the quantity to be recovered.

Linear Inverse Problems See Learning theory

Ankur Moitra : Linear Inverse Problems

Matrix completion

intro

Fill in the entries in a matrix, given a small number of observed entries.

To tackle this problem, we assume, that the matrix is approximately of low rank. This is a way to understand why it may be approximately low rank

There is a natural non-convex optimization problem to solve this

See Convex optimization heuristics for linear inverse problems for the general theory of this kind of convex relaxations