Related:
- Recovery from Linear Measurements with Complexity-Matching Universal Signal Estimation - implementation -
- Complexity-Matching Universal Signal Estimation in Compressed Sensing
- Beyond Sparsity: Minimum Complexity Pursuit for Universal Compressed Sensing
- A Low Complexity Solver ?
- Minimum Complexity Pursuit
- A Q&A with Marco Duarte and Dror Baron
Liked this entry ? subscribe to Nuit Blanche's feed, there's more where that came from. You can also subscribe to Nuit Blanche by Email, explore the Big Picture in Compressive Sensing or the Matrix Factorization Jungle and join the conversations on compressive sensing, advanced matrix factorization and calibration issues on Linkedin.
1 comment:
Igor, thanks for sharing with the community without my even pointing it out. Your ability to retrieve information is always impressive.
I'd like to mention to the audience that our other video on universal denoising and approximate message passing (AMP-UD algorithm) may be the one you prefer to watch. The reason is that the Monte Carlo (MCMC) approach is slower, and can't reach the minimum mean square error (MMSE), which is the best-possible reconstruction performance that one might hope for (I'm assuming that you're interested in square error; for unconventional types of error, see Jin Tan's video from 2013).
In contrast, AMP-UD is reasonably fast (signals of length 10,000 are processed in under 5 minutes), it gets really close to the MMSE for i.i.d. inputs that we've looked at, and on non-i.i.d. inputs - including real data - it usually reconstructs better than the MCMC approach shown in this video.
Feel free to watch both. If you like them, we'll keep making them!
Post a Comment