June 2025 - Random - Mailing list — Dipartimento di Matematica, UniPI

Seminario CNR-IAC
by GIOVANNI LUCA TORRISI 24 Jun '25

24 Jun '25

Oggi, martedì 24 giugno 2025 alle ore 14:30 Dario Trevisan, Dipartimento di Matematica , Università di Pisa , terrà il seminario: Convergenza quantitativa delle reti neurali profonde ai processi gaussiani Astratto In questo seminario, esploriamo la convergenza quantitativa di ampie reti neurali profonde con pesi gaussiani a processi gaussiani, stabilendo nuove velocità per la loro approssimazione gaussiana. Mostriamo che la distanza di Wasserstein tra l'output della rete e la sua controparte gaussiana scala inversamente alla larghezza della rete, con limiti applicabili a qualsiasi insieme finito di input in specifiche condizioni di non degenerazione delle covarianze. Inoltre, estendiamo la nostra analisi al framework bayesiano, studiando i posteriori esatti per le reti neurali, quando dotate di priori gaussiani e funzioni di verosimiglianza regolari, ma forniamo anche recenti progressi nell'approssimazione quantitativa di reti addestrate tramite discesa del gradiente nel regime NTK. Basato su lavori congiunti con A. Basteri, A. Agazzi ed E. Mosig. Il seminario si svolgerà presso la sede CNR-IAC di Roma , sala al primo piano , in modalità mista: in presenza e in streaming al link https://www.youtube.com/watch?v=RMKVBN5l13g Cordialmente, Giovanni Luca Torrisi

1 0

Seminari Giorgio Cipolloni - Politecnico di Milano - 24 e 25 Giugno 2025
by Luca Scarpa 23 Jun '25

23 Jun '25

Si segnalano i seguente seminario a tutti gli interessati. ----------------------------------------------------------------------------------------------------- Martedì 24 Giugno 2025, ore 15:00. Aula Seminari III Piano, Dipartimento di Matematica, Politecnico di Milano. Speaker: Giorgio Cipolloni, The University of Arizona, College of Mathematics. Title: Spectral Statistics of Large Random Matrices Abstract: The last fifteen years have brought major breakthroughs in understanding the spectra of large random matrices. I'll discuss several recent advances in eigenvalue and eigenvector statistics for both Hermitian and non-Hermitian matrices, focusing on the universal patterns that emerge. Link Zoom: https://polimi-it.zoom.us/j/92959315270?pwd=heQYe7W7vKv2UUf8mmT1mRuyW57F9B.1 Link Seminario Polimi: https://www.mate.polimi.it/eventi/?id=2597#single ----------------------------------------------------------------------------------------------------- Mercoledì 25 Giugno 2025, ore 15:00. Aula Seminari III Piano, Dipartimento di Matematica, Politecnico di Milano. Speaker: Giorgio Cipolloni, The University of Arizona, College of Mathematics. Title: Logarithmically Correlated Fields from Random Matrices. Abstract: In 2012, Fyodorov, Hiary, and Keating discovered a new connection between random matrices and extremal values of logarithmically correlated fields, such as branching Brownian motion and the 2D Gaussian Free Field. In particular, they conjectured that extreme statistics of random characteristic polynomials of unitary matrices belong to the universality class of logarithmically correlated fields, which was first identified by Bramson in his seminal work on branching Brownian motion. I will review several contributions in proving this connection and present some new recent works establishing new connections of characteristic polynomials of non-Hermitian matrices and two- and three-dimensional logarithmically correlated fields. Link Zoom: https://polimi-it.zoom.us/j/92092890590?pwd=81rN5Wr71qo5hPv5UkGhkm5MdPM7Rv.1 Link Seminario Polimi: https://www.mate.polimi.it/Eventi/?id=2598#single ----- Prof. Luca Scarpa, PhD Associate Professor in Probability Department of Mathematics Politecnico di Milano Via E. Bonardi 9 20133 Milano, Italy e.mail: luca.scarpa(a)polimi.it<mailto:luca.scarpa@polimi.it> url: https://sites.google.com/view/lucascarpa

1 0

Prof Scalas Seminar Sant'Anna Pisa, palazzo Boyl
by Francesco Montomoli 19 Jun '25

19 Jun '25

Ciao a tutti Oggi pomeriggio ore 15, Prof Scalas, Sapienza, tiene un seminario Palazzo Boyl, aula dottorandi, secondo piano CISC su Uncertainty Quantification con code grasse Siete benvenuti A presto Francesco

1 0

Final Reminder Eurandom Workshop on Dependence Modeling (NL)
by Elisa Perrone 19 Jun '25

19 Jun '25

Dear Colleagues, This is a final reminder to register for the upcoming Eurandom Workshop on Dependence Modeling, taking place on June 30 - July 2, 2025, at the Eindhoven University of Technology (NL). Workshop website and program: Eurandom Workshop on Dependence Modeling <https://www.eurandom.tue.nl/event/https-www-eurandom-tue-nl-event-yes-depen…> Registration Deadline: Sunday, June 22. If you plan to attend and haven’t yet registered, we encourage you to do so as soon as possible. Special Journal Issue: We are pleased to announce a special issue of the journal Dependence Modeling <https://www.degruyterbrill.com/journal/key/demo/html?lang=en&srsltid=AfmBOo…>, exclusively open to workshop participants. This is a great opportunity to contribute to the field and showcase your work developed in connection with the workshop themes. Sponsors: This workshop is generously sponsored by Eurandom, NWO, 4TU.AMI, and STAR. We thank them for their support in fostering research collaboration and advancements in dependence modeling. We look forward to welcoming you in Eindhoven! Best wishes, Elisa Perrone (on behalf of all the organizers) --- Elisa Perrone, Ph.D. Assistant Professor Department of Mathematics and Computer Science Eindhoven University of Technology elisaperrone.info <http://elisaperrone.info/>

1 0

I: [owps] Fwd: next OWPS: Eva Löcherbach and Elisa Marini
by Luisa Andreis 18 Jun '25

18 Jun '25

Dear colleagues, this is a gentle reminder of today One World Probability Seminar, details below. The seminar will be held at 15:00 (italian time). This will be the last seminar of the spring term, we hope to see many of you online! Luisa and Roger ________________________________ Da: owps-request(a)lists.bath.ac.uk <owps-request(a)lists.bath.ac.uk> per conto di One World Probability <ow.probability(a)gmail.com> Inviato: martedì 17 giugno 2025 21:13 A: owps(a)lists.bath.ac.uk <owps(a)lists.bath.ac.uk> Oggetto: [owps] Fwd: next OWPS: Eva Löcherbach and Elisa Marini The final OWPS of Spring will be on Wednesday, June 18, from 13:00 to 15:00 UTC time. Title, abstract and the zoom link are below the signature and can be found on the website https://www.owprobability.org/one-world-probability-seminar<https://protect-eu.mimecast.com/s/-zGkCWqjZFlpkVlsnEyR_?domain=eur01.safeli…>. ----------------------- Probabilistic models for systems of interacting spiking neurons and some words about their mean field limits E. Löcherbach (Paris 1 Panthéon-Sorbonne University) I will give an overview of recent results about mean field limits for systems of interacting point processes modeling spiking (biological) neurons. I will start with a short introduction to the functioning of neurons and the modeling of their spiking activity by stochastic integrate and fire models and then focus on a particular class of models which are stochastic intensity based processes. Then we will discuss mean field limits and propagation of chaos results for large homogeneous systems of neurons and see how the limit is described by a McKean-Vlasov type equation driven by Poisson random measure. A second part of the talk is devoted to the study of systems with random synaptic weights in a diffusive scaling. We will see how this setting leads to conditional propagation of chaos and how the convergence can be obtained by means of a new martingale problem. Finally, if time permits, I will also discuss the longtime behavior both of the finite and the limit system of neurons. The second part of the talk is based on joint work with Xavier Erny and Dasha Loukianova. Strong conditional propagation of chaos for systems of interacting particles with nearly stable jumps E. Marini (Dauphine-PSL Paris) We consider a system of N interacting particles, described by SDEs driven by Poisson random measures, where the coefficients depend on the empirical measure of the system. Every particle jumps with a rate depending on its position. When this happens, all the other particles of the system receive the same random kick, distributed according to a heavy-tailed random variable belonging to the domain of attraction of an α-stable law, and scaled by N−1/α, where α ∈ (0, 2) \ {1}. We call these jumps collateral jumps. Moreover, in the case 0 < α < 1, the jumping particle itself undergoes a macroscopic, main jump. Similar systems are used to model networks of interacting neurons; in that context, main and collateral jumps represent, respectively, the hyperpolarization of a neuron after a spike and the synaptic inputs received by post-synaptic neurons from pre-synaptic ones. We prove that our system satisfies the conditional propagation of chaos property: as N → +∞, the finite particle system converges to an infinite exchangeable system that obeys a McKean–Vlasov SDE driven by an α-stable process. In the limit, the particles become independent, conditionally on the driving α-stable process. https://polimi-it.zoom.us/j/92945513591?pwd=zjtRwpHoO9kRyQuPPj4o186jXrvg1v.1 Meeting ID: 92945513591 Passcode: 131676

1 0

CLADAG 2025: extended deadlines and call for special issues
by Marta Nai Ruscone 17 Jun '25

17 Jun '25

*** apologies for cross-posting *** Dear All, in reminding you that the deadline for abstract submission to CLADAG-VOC 2025 is next July 1st, we are pleased to inform that the call for Special Issues of ADAC and METRON related to next CLADAG-VOC 2025 are ready (detailed information are here <https://urlsand.esvalabs.com/?u=https%3A%2F%2Fcladag2025.unina.it%2Fsi%2F&e…>.). The journal Advances in Data Analysis and Classification (https://link.springer.com/journal/11634 <https://urlsand.esvalabs.com/?u=https%3A%2F%2Flink.springer.com%2Fjournal%2…>) will publish a Special Issue on Data Modelling for Dimension Reduction, Clustering and Classification with Complex Data. Guest editors are Mark de Rooij (Leiden University, the Netherlands) Michele La Rocca (University of Salerno, Italy) Andrea Cerioli (University of Parma, Italy) Important dates: November 30th, 2025: Submission of full papers for the Special Issue; April 1st, 2026: Notification to Authors; June 30, 2026: Final papers. More information will be soon available on the panel “Journal updates” in the journal website <https://urlsand.esvalabs.com/?u=https%3A%2F%2Flink.springer.com%2Fjournal%2…>. The journal METRON (http://www.springer.com/statistics/journal/40300 <https://urlsand.esvalabs.com/?u=http%3A%2F%2Fwww.springer.com%2Fstatistics%…>) will publish a Special Issue on Theoretical and Computational Advances in Classification, Clustering, Pattern Recognition and Data Analysis. Guest editors are: Carlo Cavicchia (Erasmus University Rotterdam, the Netherlands) Antonio D'Ambrosio (University of Naples Federico II, Italy) Antonio Punzo (University of Catania, Italy) Important dates: November 30th, 2025: Submission of full papers for the Special Issue; February 28, 2026: Notification to Authors; May 31, 2026: Deadline for submitting the revised paper September 30, 2026: Publication of the Special Issue (tentative). More information are available on the dedicated web page of the journal <https://urlsand.esvalabs.com/?u=https%3A%2F%2Flink.springer.com%2Fjournal%2…>. Detailed information about next CLADAG-VOC 2025 are available here <https://urlsand.esvalabs.com/?u=https%3A%2F%2Fcladag2025.unina.it%2F&e=ed7a…>. Kind regards, Antonio D'Ambrosio Michele La Rocca Mark de Rooij

1 0

Seminario Bin Yu al dipartimento G. Catelnuovo
by Gustavo Posta 17 Jun '25

17 Jun '25

Lunedì 23 giugno alle 16:00 nella Sala di Consiglio il prof. Bin Yu (Statistics, EECS, Center for Computational Biology, and Simons Institute UC Berkeley) terrà un seminario dal titolo: Sparse dictionary learning and deep learning in practice and theory Abstract: Sparse dictionary learning has a long history and produces wavelet-like filters when fed with natural image patches, corresponding to the V1 primary visual cortex of the human brain. Wavelets as local Fourier Transforms are interpretable in physical sciences and beyond. In this talk, we will first describe adaptive wavelet distillation (AWD) to turn black-box deep learning models interpretable in cosmology and cellular biology problems while improving predictive performance. Then we present theoretical results that, under simple sparse dictionary models, gradient descent in auto-encoder fitting converges to one point on a manifold of global minima, and which minimum depends on the batch size. In particular, we show that when using a small batch-size as in stochastic gradient descent (SGD) a qualitatively different type of â€œfeature selectionâ€ occurs. Gli interessati sono invitati a partecipare. ---------------------------------------------------------------- Gustavo Posta Dipartimento di Matematica Università di Roma "la Sapienza" P.le A. Moro 2, 00185 Roma Italy web: http://www1.mat.uniroma1.it/~posta e-mail: gustavo.posta(a)uniroma1.it phone: +39-06-4991-4969 ----------------------------------------------------------------- -- *Fai crescere le giovani ricercatrici e i giovani ricercatori*** *con il 5 per mille alla Sapienza* Scrivi il codice fiscale dell'Università *80209930587 **Cinque per mille <https://www.uniroma1.it/it/node/23149>*

1 0

OWABI@BayesComp - 19th June 1pm UK time
by Massimiliano Tamborrino 16 Jun '25

16 Jun '25

Dear All, The final OWABI seminar www.warwick.ac.uk/owabi<http://www.warwick.ac.uk/owabi> of the season will feature two selected talks and will take place live from BayesComp https://bayescomp2025.sg/! The talks will take place on Thursday the 19th June at 1pm UK time (note the different date and time!), and will be given by * Filippo Pagani (University of Warwick): Approximate Bayesian Fusion * Maurizio Filippone (KAUST): GANs Secretly Perform Approximate Bayesian Model Selection with abstracts reported below. The talks will be live streamed from the following link: https://monash.zoom.us/j/81050994376?pwd=aF52Ab2oUlNYPO251d4nDw0jXRemia.1<https://www.google.com/url?q=https%3A%2F%2Fmonash.zoom.us%2Fj%2F81050994376…> Meeting ID: 810 5099 4376 Passcode: 137607 The session will run in a hybrid format, taking place live from BayesCom2025 - 8pm-9pm (Singapore time) - and will be located in room LT50 in the conference venue. We look forward to seeing you online or in person! Best, Massimiliano on the behalf of the OWABI Organisers 1st OWABI Talk: 1-1.30pm Uk time Speaker: Filippo Pagani<https://filippopagani.github.io/> (University of Warwick) Title: Approximate Bayesian Fusion Abstract: Bayesian Fusion is a powerful approach that enables distributed inference while maintaining exactness. However, the approach is computationally expensive. In this work, we propose a novel method that incorporates numerical approximations to alleviate the most computationally expensive steps, thereby achieving substantial reductions in runtime. Our approach retains the flexibility to approximate the target posterior distribution to an arbitrary degree of accuracy, and is scalable with respect to both the size of the dataset and the number of computational cores. Our method offers a practical and efficient alternative for large-scale Bayesian inference in distributed environments. 2nd OWABI Talk: 1.30-2pm Uk time Speaker: Maurizio Filippone<https://www.kaust.edu.sa/en/study/faculty/maurizio-filippone> (KAUST) Title: GANs Secretly Perform Approximate Bayesian Model Selection Abstract: Generative Adversarial Networks (GANs) are popular models achieving impressive performance in various generative modeling tasks. In this work, we aim at explaining the undeniable success of GANs by interpreting them as probabilistic generative models. In this view, GANs transform a distribution over latent variables Z into a distribution over inputs X through a function parameterized by a neural network, which is usually referred to as the generator. This probabilistic interpretation enables us to cast the GAN adversarial-style optimization as a proxy for marginal likelihood optimization. More specifically, it is possible to show that marginal likelihood maximization with respect to model parameters is equivalent to the minimization of the Kullback-Leibler (KL) divergence between the true data generating distribution and the one modeled by the GAN. By replacing the KL divergence with other divergences and integral probability metrics we obtain popular variants of GANs such as f-GANs, Wasserstein-GANs, and Maximum Mean Discrepancy (MMD)-GANs. This connection has profound implications because of the desirable properties associated with marginal likelihood optimization, such as (i) lack of overfitting, which explains the success of GANs, and (ii) allowing for model selection, which opens to the possibility of obtaining parsimonious generators through architecture search. ------ Dr. Massimiliano Tamborrino Reader (Associate Professor) and WIHEA Fellow Department of Statistics University of Warwick https://warwick.ac.uk/tamborrino

1 0

Seminars in Statistics at Collegio Carlo Alberto - Steven CAMPBELL (COLUMBIA UNIVERSITY)
by Matteo Giordano 16 Jun '25

16 Jun '25

SEMINARS IN STATISTICS @ COLLEGIO CARLO ALBERTO <https://www.google.com/url?q=https://www.carloalberto.org/events/category/s…> Venerdì 20/06/2025, presso il Collegio Carlo Alberto, in Piazza Arbarello 8, Torino, si terrà il seguente seminario: ------------------------------------------------ 12.00-13.00 Speaker: Steven CAMPBELL (COLUMBIA UNIVERSITY) Title: A Tractable Triple Problem in Sequential Inference: Filtering, Control, and Stopping Abstract: We study a Bayesian sequential inference problem for the drift of a Brownian motion in which the observer not only chooses their stopping and decision rules, but also controls the rate at which information is acquired. This extension introduces a cost for information acquisition, transforming the classical formulation of Shiryaev (1967) into a joint problem of filtering, control, and optimal stopping. Problems of this kind arise in adaptive experimental design and real-time inference across domains such as healthcare, finance, and engineering, where information is costly and decision-making must be timely. Our framework accommodates a host of classification losses and general running costs on the control process. Despite the added complexity, the problem admits a remarkably simple and semi-explicit solution that exhibits structural parallels with the classical setting. This work highlights a rare instance in which a triple problem—combining inference, control, and stopping—remains analytically tractable. Based on joint work with Georgy Gaitsgori and Richard Groenewald. ------------------------------------------------ Sarà possibile seguire il seminario anche in streaming: chiunque volesse collegarsi è pregato di inviare una email entro *mercoledì 18/06/2025* a matteo.giordano(a)unito.it <mailto:matteo.giordano@unito.it> Il webinar è organizzato dalla "de Castro" Statistics Initiative (www.carloalberto.org/stats <http://www.carloalberto.org/stats>) in collaborazione con il Collegio Carlo Alberto. Cordiali saluti, Matteo Giordano Assistant Professor (RTDA) Department of Economics, Social Studies, Applied Mathematics and Statistics (ESOMAS) www.matteogiordano.weebly.com <https://matteogiordano.weebly.com/>

1 0

Seminario Padova - Filippo Quattrocchi
by alekos.cecchin＠unipd.it 15 Jun '25

15 Jun '25

Buongiorno a tutti, Vorremmo segnalarvi che venerdì prossimo (20 Giugno) in aula 2BC30 (Torre Archimede, Università di Padova) ci sarà un seminario per il ciclo di seminari in Probabilità e Finanza di: Filippo Quattrocchi (Institute of Science and Technology Austria (ISTA)) <https://sites.google.com/view/filippoquattrocchi/> https://sites.google.com/view/filippoquattrocchi/ Title: Acceleration-based Optimal Transport Date: June 20, 2025, at 14:30, 2BC30 Abstract: Finding smooth interpolations between probability measures is a problem of broad interest, with applications ranging from physics to biology, from computer graphics to engineering. In this talk, I will discuss a model in which such interpolations are found by minimising an action functional of the acceleration, and present recent results obtained in collaboration with G. Brigati and J. Maas (ISTA), arXiv:2502.15665. The first result is a fluid-dynamic equivalent formulation of the model, previously conjectured by Y. Chen, G. Conforti, and T. T. Georgiou. We complete the formal argument outlined by the aforementioned authors through the introduction of two new mathematical ingredients. The second result is the description of the (hypoelliptic) Riemannian structure induced by the minimal acceleration-based action on the space of probability measures. Vi aspettiamo numerosi! Alberto Chiarini e Alekos Cecchin Sito web del seminario: https://www.math.unipd.it/~chiarini/seminars/

1 0