𝗔𝘆𝘂𝘀𝗵

On the SIAM News Blog, Ayush Bhandari explains that the unlimited sensing framework improves upon traditional #digitization by treating #quantization noise errors as data and saving bits by folding signals instead of clipping them. Read more here! siam.org/publications/s…

@mmbronstein Spectacular!

Nepal from above

Great to have Giovanni at @imperialeee today, who gave an amazing talk on inverse problems. And hopefully first of more visits to come... !

@cesifoti Got to join your group, César!

The Center for Collective Learning (CCL) has grown enormously this year. Here we are ready to enter the laser tag arena in Budapest a few weeks ago :-)

New Tutorial on 𝗨𝗻𝗹𝗶𝗺𝗶𝘁𝗲𝗱 𝗦𝗲𝗻𝘀𝗶𝗻𝗴 at the upcoming @ieeeICASSP w/ @Ruiming999. We'll cover the pet peeves of @IEEEsps: • Sparse + sub-Nyquist Recovery • Array Sig. Proc. • Computatnal Imging • Communication Sys. etc. from the lens of Modulo Non-linearities.

@wicko3 @imperialeee John, thank you for the perfect organization!

Delighted to have our first “Meet the Profs” seminar of the term, featuring Ayush Bhandari (@AB2World) who showed us how to extract a useful signal just from the noise of quantisation! Simultaneously mind-blowing and beautifully explained. @imperialeee

The kind of things you stumble upon in London …

Feels like a passing of the torch between fields. When I was a teenager in the 1980s, after a half-century of monumental progress powered by theoretical physics, most smart high schoolers wanted to do physics. Upon arriving as an undergrad at MIT in 1988, it quickly became clear that the next big thing was computer science. Congratulations to @geoffreyhinton and John on this honor. Interestingly, I met John Hopfield a long time before I met Geoff. ( In mid 1990s when he invited me to give a guest lecture in his Princeton ugrad class about my algorithm for traveling salesman. )

𝗣𝗿𝗼𝗻𝘆'𝘀 𝗠𝗲𝘁𝗵𝗼𝗱 is a widely used, 200 years old technique. It estimates unknown parameters of K sine-waves from 2K samples. In our new work (IEEE TSP, @IEEEsps) on Unlimited Sensing, we show that same can done with 6K+4 non-linear, modulo samples. arXiv: 2409.16472

@mmbronstein Wow! That’s amazing, Michael!

@salmanasif @ERC_Research @ImpEngineering Salman, thank you!

@AB2World @ERC_Research @ImpEngineering Awesome, congrats!

Thank you @ERC_Research. Our idea is simple: Use 𝗤𝘂𝗮𝗻𝘁𝗶𝘇𝗮𝘁𝗶𝗼𝗻 𝗡𝗼𝗶𝘀𝗲 as the signal. The gratifying part is that UG/MS students (@ImpEngineering) can get this and have already contributed to publications. Here's an overview of our scientific vision.

@hijizhou @ERC_Research @ImpEngineering Thank you, Jizhou! (somehow missed your text)

@AB2World @ERC_Research @ImpEngineering That's cool!

@taiyasaki @ERC_Research @ImpEngineering Here's an example to contextualize the idea + link to a paper with an algorithm that shows hardware experiments. youtu.be/prV40WlzHh4

@taiyasaki @ERC_Research @ImpEngineering Andrea, thank you! The key idea is: any signal = integer + fractional part. We have already developed hardware that directly captures the frac-part (~quantization noise). Signals are recovered algorithmically. Interestingly, this can be done at Nyqiust-rate (const. factor).

@WeisiG Thank you. This is very relevant. We have developed hardware that can directly capture "quantization noise" which is just the fractional-part of the signal. I hope this gives you some idea. youtu.be/I9f1S0F3s5E

@AB2World Mega cool, 🙋🏻‍♂️ can I ask a stupid question. How do you know what the quantisation noise is if you don’t have the original signal to compare with after ADC?

@IlyasHairline @ERC_Research @ImpEngineering Pls. be my guest!

@AB2World @ERC_Research @ImpEngineering I think I can use this

@thephotonguy @ERC_Research @ImpEngineering Thanks, Anchit!

@AB2World @ERC_Research @ImpEngineering Congratulations!

@AjdDavison @ERC_Research @ImpEngineering Thank you, Andrew!

@AB2World @ERC_Research @ImpEngineering Congratulations Ayush!

@yaniv_romano @ERC_Research @ImpEngineering I'll use your words here, "likewise." Very happy for you!

@AB2World @ERC_Research @ImpEngineering Amazing!! Congratulations :)

@laurentduval @ERC_Research @ImpEngineering I hope so, Laurent. Thank you!

@AB2World @ERC_Research @ImpEngineering Amazing science to come, walking