
Professor of Mathematical Analysis @malga_center, @UniGenova. Previously at ETH Zurich, École Normale Supérieure Paris, Oxford University.
Genoa, Italy
Joined July 2012
- Tweets893
- Following307
- Followers381
- Likes781
Giovanni S. Alberti retweeted
Diffusion models have a fundamental flaw: when trained on real-world noisy data, they just learn to perfectly generate that exact noise. We found a way to fix it. Introducing Manifold Attracted Diffusion (MAD), accepted at ICML 2026 🧵
📢 I’m happy to share that an interview about my research and academic journey has just been published in the @euromathsoc Magazine!
This interview is part of a series in which @ERC_Research grantees are asked about their experiences and research.
ems.press/journals/mag/artic…
1
2
130
We discussed various aspects of my work, with a special focus on my ERC project, as well as some broader reflections on applied mathematics and academic training.
Many thanks to @mt_wolfram and Marc E. Pfetsch for a very engaging discussion!
95
I will give a talk on constraints for solutions to PDEs at the International Zoom Inverse Problems Seminar on
📅 Thursday 9th
⌚️ 6pm CEST, 12pm EDT, 9am PDT
sites.uci.edu/inverse/
100
I am very happy to share that this work has just been published in SIAM Journal on Applied Mathematics!
epubs.siam.org/eprint/EKHRVY…
New paper on #machinelearning for #inverseproblems: we show that sampling methods for inclusion detection are neural networks, and can be learned!
Joint work with Shiwei Sun
@malga_center, @UniGenova
6
248
New preprint "A convex lifting approach for the Calderón problem", by @GSAlberti, @rom1petit and @SimoneSanna1893
By lifting the product q(x)u(x) as a rank one operator, it is possible to write the inverse problem as a linear problem. We show that, if a non-degenerate source condition holds, the proposed convex problem has a unique solution that is the sought-after coefficient.
1
148
Giovanni S. Alberti retweeted
MaLGa - Machine Learning Genoa Center ospita un seminario di @TomasoPoggio per ripercorrere le tappe fondamentali della ricerca sull'#intelligenzaartificiale da una prospettiva personale.
📌10 marzo 2025, ore 15:00, Genova
ℹ️unige.it/news/tomaso-poggio-…
#UniGe #UniGenova #TomasoPoggio
Giovanni S. Alberti retweeted
🧵 Excited to share our new preprint on learning optimal regularizers for inverse problems! With @GSAlberti, EDeVito, THelin, MLassas, LRatti
Giovanni S. Alberti retweeted
Giovanni S. Alberti, Alessandro Felisi, Matteo Santacesaria, S. Ivan Trapasso: Compressed sensing for inverse problems II: applications to deconvolution, source recovery, and MRI arxiv.org/abs/2501.01929 arxiv.org/pdf/2501.01929
In effetti, un'intelligenza artificiale affidabile deve saper dire "Non lo so"
arxiv.org/abs/2408.02357
Replying to @RobertoBurioni
PS: ho fatto una cosa semplice, ho posto una domanda difficile a Grok. Mi ha risposto in modo errato, ma credibile. Ho detto che era sbagliato, ha fornito una seconda risposta credibileed errata. Una persona intelligente mi avrebbe risposto subito "non lo so".
2
156
Talk on super-resolution in inverse scattering, with @rom1petit and @matsanta
Slides: giovannisalberti.github.io/t…
Funded by @ERC_Research
@malga_center @UniGenova
Today I am giving an online talk on learning (simple) regularizers for inverse problems for the Data-Enabled Science Seminar, @UHouston
18:45 CET, 11:45 local time
Slides: giovannisalberti.github.io/t…
1
104
Giovanni S. Alberti retweeted
📅Today 5 PM
📍Accademia Ligure di Scienze e Lettere
🎙@AgneseSeminara @UniGenova
"#AI to Decode Animal Behavior in Unpredictable Physical Environments" (Italian🇮🇹) explores how #reinforcementlearning helps understand how animals navigate turbulent environments using sensory cues













