CS: Boogie Woogie Grid, Compressive sensing (CS) and sparse representations (SR) for computer vision and image processing and a Postdoc.

First, happy new year to all. 

In a recent entry, I was looking for an algorithm that could produce a Boogie Woogie grid. I then asked a similar question on Math Overflow and got the following to this article The Mondrian Process by Daniel M. Roy and Yee Whye Teh. One video of this work is shown above. It looks great, I now need an implementation of that.

I just found a talk Compressive sensing (CS) and sparse representations (SR) for computer vision and image processing by Rama Chellappa. I don't know if this is because of my ISP,, but anytime I need to see something from Microsoft Research, I need to go through a proxy :-). The presentation itself is very much focused on the group of researchers featured and their work. I don't think it can be construed as an overview of the whole field as it relates to computer vision and image processing, however it is a good start.

Laurent Demanet tweeted about a job in his group:

Postdoctoral Researcher in Computational Mathematics

The Imaging and Computing group in the Department of Mathematics at MIT invites applications for two postdoctoral positions in computational mathematics, with focus on some subset of the following areas: computational wave propagation, optimization, inverse problems, applied harmonic analysis, sparsity (compressive sensing), fast algorithms, radar imaging, seismic imaging. These positions are part of related efforts to 1) design the next generation of multiscale transforms in the context of high-dimensional datasets, 2) investigate new implications of sparsity ideas to inverse problems, and 3) research new ideas for solving computational wave propagation problems in the high-frequency regime (mostly Helmholtz). Applications to radar imaging or seismic imaging will be considered. Candidates are expected to have a strong background in computational mathematics and should be very comfortable with one or more of the areas mentioned above. Programming experience in C, C++, Fortran, MPI, or CUDA/OpenCL is a plus. The balance of work between theoretical and computational will depend on the candidate's affinities. The position may be offered for 1, 2, or 3 years. It may involve some teaching within the Mathematics department as part of an lecturer/instructor position. The salary is competitive. The position is expected to start in the summer or fall of 2011, but there is flexibility.

The application file should contain a CV, a list of publications, and a research statement. The research statement should, in no more than a few pages, summarize the candidate's past research accomplishments and vision for possible future research. Follow either of the following two options to apply for the position:

a) apply via mathjobs.org through MIT's instructor position, and sent an email to Prof. Laurent Demanet to express your interest in the postdoctoral position;

b) or alternatively, the application package should be emailed (not mailed) to Prof. Laurent Demanet. The candidate's message should contain "Postdoc position" in the subject. In addition, the candidate should arrange for at least three recommendation letters to be sent to the same email address. (It is not enough to list references in your resume. The letters should be sent directly by the referees and should not be seen by the candidate.)

Screening of applications will begin on December 1st, 2010, and will continue until the positions are filled. MIT is an equal-opportunity, affirmative-action employer.