Matching Pursuit¶
Matching pursuit and related greedy algorithms iteratively select sources that best explain the residual signal. They produce sparse solutions and can handle correlated sources effectively.
This category contains 6 solvers.
Solvers¶
| Full Name | Solver ID | Description |
|---|---|---|
| Compressive Sampling Matching Pursuit | CoSaMP |
Greedy sparse recovery that identifies 2K candidate atoms (singletons or patches) using all time points jointly (MMV)... |
| Iterative Subspace Smooth Matching Pursuit | ISubSMP |
Smooth, subspace-based matching pursuit variant that operates on a spatially smoothed (patch) dictionary and iterates... |
| Orthogonal Matching Pursuit | OMP |
Greedy sparse recovery that iteratively selects the best-correlating atom (singleton or spatial patch) using all time... |
| Reduce Multi-Measurement-Vector and Boost | ReMBo |
Randomly reduces a multi-measurement problem to repeated single-measurement OMP-style recovery with patch-dictionary ... |
| Subspace Pursuit | SP |
Greedy sparse recovery that alternates between support expansion and pruning (singletons or patches) using all time p... |
| Subspace Smooth Matching Pursuit | SubSMP |
Smooth matching pursuit variant that projects data into a low-dimensional subspace and merges patch supports across c... |