sparse representation is:
A signal can be represented as a linear combination of basis functions where the set of basis functions is called dictionary and data samples are much more than their features. Mathematically, in the system of linear equations $Y=DX$ where $Y \in \mathbb{R}^{n \times N} (n \ll N)$ we seek a dictionary that results in sparse representation of $Y$.

Digging the history of sparse representation, my take is it all started by using transforms as D; wavelet, Fourier and such. They were good in the application they were designed for but a generalization to a wider area was a failure. So a combination of basis functions was used as a dictionary D, an overcomplete one with more basis functions than the size of each basis function. There are lots of examples of different transforms and their result. But I cannot think of an example for an overcomplete dictionary. Is it due to the fact that they are learned and there is no special dictionary to be used? I mean, given an initial dictionary, it is learned in an iterative process to better fit the data, is there a well-known instance of such a dictionary like the case JPEG2000 using wavelet packets as D?



Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Browse other questions tagged or ask your own question.