Ask a Question

Prefer a chat interface with context about you and your work?

Learning to solve TV regularised problems with unrolled algorithms

Learning to solve TV regularised problems with unrolled algorithms

Total Variation (TV) is a popular regularization strategy that promotes piece-wise constant signals by constraining the l1-norm of the first order derivative of the estimated signal. The resulting optimization problem is usually solved using iterative algorithms such as proximal gradient descent, primal-dual algorithms or ADMM. However, such methods can require …