Challenging Datasets

Recently, at SPARS11, by featuring the example of the work of Miki Lustig at Berkeley,  David Donoho reminded us of the virtue of doing with what you had as a constraint to push the envelope of the algorithms capabilities and extract more information than conventional techniques. I could not escape trying to make a parallel between this and the need to look at older datasets and see if something better could come out of them. Following up on Leonardo's heart entry, I eventually got in touch with Tom Kayes, Robert Crawford and Art Andersen, three interesting gentlemen. There is not much to talk about for the moment as we are still at the getting to know each other stage.

One thing for sure is that instead of having one CT reconstruction problem with few views for a dinosaur that died a hundred million years ago, it seems that some of the best reconstruction algorithms currently developed by many of you could take a stab at two different problems. The second one being about a meteorite reentry. Both datasets are clearly outside the usual datasets used by y'all. My point to Tom, Robert and Art  has revolved around making available some of their truly unique datasets to the scientific community so that they can eventually be used as benchmarks. Clearly for different reasons, these datasets are extremely difficult to obtain in the first place as the radiation dose for the Dino was pretty high and the meteorites are pretty far (but not point like).They are good examples of the type of data that can only push the envelope of some of the algorithms we have. Stay tuned, we'll see how this goes. 

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.