Monday, March 01, 2010

CS: An update featuring SPAMS, chaos, poolMC and the DNA Sudoku video

A while back, I had made a request to Julien Mairal and his co-authors about the need for their potentially very fast online dictionary learning toolbox to be available for win32 systems as some of us do not do Linux/Unix. Julien came through with his promise to look into it and sent me the following:

Dear Igor,

I have just released a Windows 32bits version of the SPAMS toolbox, which is available at I hope that it works ok, I could not test it intensively.

Best regards


Thank you Julien !

Laurent Duval sent me the following notice for a presentation in french on the use of a chaotic system to build compressed sensing measurement matrices. It will be featured at the french speaking 13e Rencontre du Non Lineaire (10-12 mars 2010) at ESPCI. The presentation is entitled `Compressive Sensing' en utilisant le Chaos par Lei Yu, Jean-Pierre Barbot, Gang Zheng and Hong Sun, présenté par Jean-Pierre Barbot. The abstract reads:
La methode dite de l'acquisition comprimee plus connue sous le vocable anglo-saxon de `Compressive Sensing' est une nouvelle methode qui permet de capturer et de retrouver par la suite un signal echantillonne a des frequences sous Nyquist. Afin de garantir la reconstitution parfaite du signal cette methode requiere la construction d'une matrice dite `sensing' matrice possedant des proprietes d'inversion particulieres. Ici, une construction de cette matrice a l'aide de sequences issues d'un systeme chaotique est proposee et il est prouve que cette matrice verifie avec une ecrasante probabilite (superieure a une construction aleatoire de type Gaussien) les proprietes de reconstruction requises.

Thanks Laurent for the tip!

Raghu Kainkaryam reminded me of how much time I personally waited to see the poolMC paper with the clear impact it will have on experimental pooling design and how it connects to compressive sensing:

Hi Igor,

Thanks for linking to poolMC. It's been a long time coming. Hope to get some valuable feedback from the CS community.

With regards,

The paper was featured in last week's longest entry. Check it out. If you recall there was a video by Anna Gilbert showing some preliminary results in the compressive sensing online talks pageand in the compressive sensing hardware page. Thanks Raghu!

While we are on the subject of compressive sensing inspired pooling, here is Yaniv Erlich's video presentation of the DNA sudoku paper mentioned earlier here.

No comments: