Thursday, April 18, 2013

Nuit Blanche Readers' Review: MOUSSE, Google Reader replacement, QuantCS, Supelec

Magnificent shot of the non responsive ENVISAT by the Pleiades 1A satellite flying 100km over ENVISAT (CNES)

I was talking to Yao Xie about the MOUSSE package that was featured recently here as I could not find the simple example in the initial package to be downloaded, Yao kindly responded with:

Hi Igor,

Thanks for your feedback and suggestions! I have updated the folder for this simple parabola demo and created a readMe file. The link to it is

You could find the new readMe file in this folder explaining the files..... the solar flare example is available in a [much] bigger folder:

Thanks Yao ! as a reminder the MOUSSE site is here.

Yesterday, in Around the webs in 78 hours, I asked about replacements to Google Reader that is slated to be terminated on July 1st, here is what Bourbaki said in the comment:

After first switching to "the old Reader" which turned out to be slow and lack an offline capability, I switched over to Feedly a couple weeks ago which has Firefox and Chrome browser plug-ins as well as an iOS app (all three of which I use).
Feedly seems to work great except for offline (a refresh will make all the posts inaccessible until you are online again) and will probably be my replacement. Some tweaks to the preferences are necessary, but Feedly kindly has a blog post for incoming Reader users.
I had previously used the Newsify app which has a great offline capability (that I reverted to on a recent flight) and better interface than Feedly but they were kind of blind-sided by Google's announcement and haven't come up with an alternative yet. Newsify doesn't have a browser and sometimes failed to update the posts I read in Reader. 
Feedly is a decent all-around replacement, but I'm still looking around.
If, like Bourbaki, you have a newer set up that you feel comfortable using, please share it with us. 

In other news, Petros Boufounos made the following very interesting request: 
I hope all is well. I am wondering if there is a concise way to get all your posts about CS and quantization. I am preparing (with Laurent Jacques) slides for a tutorial on modern quantization for ICASSP, and I would like to link to your pages for further resources and reading. I found the 1-bit CS pages through the tagging system: Is there some equivalent for general quantized CS? 

Yes, Petros, all is well, considering my town hasn't been bombed lately. Since the search feature for blogs in broken on the blog, I went ahead and tagged all the entries that had the word quantized with the QuantCS tag, you can find all 52 entries that have that word here at:

This is very interesting request. If you have a similar request to Petros, please let me know and let's talk. Also if you believe that a particular entry should be labeled with a particular tag, please add that word in the comments of that entry. If there are enough of them I will make a tag out of them. 

PS: By the way, I just realized that Petros' site had an RSS capability. To follow Petros' new stuff, you can add this feed. I have done so in the right hand part of this blog.

Finally, Francesca Bassi invited me to give a talk at Supelec next week, here is the announcement:

Speaker: Igor Carron
Séminaire organisé par la division Télécoms et Réseaux
Date: Wed, 04/24/2013 - 11:00 to 12:30
Lieu: Salle du Conseil, L2S
According to some sources, in 2010, we had already generated more bits than there were stars in the Universe. These exceptionally large datasets are not within the realm of far flung scientific endeavors anymore. Very soon, fields like genomics are bound to provide even larger amounts of data that will be of direct interest to the population at large - i.e. personalized medicine -. In light of this impending constraint, here are some open questions: 
  • Do we have sensing systems that can acquire the most relevant data in a compressed fashion ? 
  • Do we have the means to making sense of that data ?
Ever since the papers of Candes, Tao, Romberg [1] and Donoho [2] nine years ago, the idea of connecting sensing and some prior information has shown to be a potential solution to some of these questions. The field has grown to be known as compressive sensing and is investigated by a diverse crowd of researchers that include mathematicians, theoretical computer scientists, statisticians, physicists all the way to engineers building sensors. In this talk, I will try to provide an overview of this very dynamic field as it has unfolded so far [3] and will point to ways one can identify compressive sensing systems in the wild. 
[1] E. J. Candès, J. Romberg and T. Tao. Robust uncertainty principles: exact signal reconstruction from highly incomplete frequency information. IEEE Trans. Inform. Theory, 52 489-509.
[2] Compressed Sensing, :Information Theory, IEEE Transactions on Information Theory, April 2006, Donoho, D.L.
Igor Carron has a Ph.D from Texas A&M University (1996) and a diplome d'ingenieur from INPG (Grenoble INP). He writes a blog on the internets called "Nuit Blanche" that is focused on compressive sensing, advanced matrix factorization and randomized numerical linear algebra. The blog has had more than 2 million page views so far and is
being read by more than a thousand researchers worldwide every day.

As some of you might have noticed, that talk has probably yielded the following blog entries:
This also led me to discover that somebody had written an "Acquisition comprimée" entry on the french wikipedia. Naturally, I had to do some minor corrections. All in all, I think both of the english and the french entry need to be re-written. All of them are focused on l_1when we know that l1 is not really about compressive sensing.

I will make the talk available on the interwebs afterwards.

Credit Photo: CNES, Pléiades 1A observes Envisat, 20 avril 2012, Le 15 avril dernier, le satellite imageur Pléiades du CNES a pris en photo à la volée le satellite Envisat de l’ESA alors qu’il passait à une distance de 100 km.

No comments: