Tomorrow...maths with oncohematology and indoor positioning! 😍😍Waiting for Sabino Ciavarella, MD, PhD Domenico Colucci Nextome

Great to organize INTRUSION2023 with Matteo Icardi, M. Putti, Fabio Difonzo, Wietse Boon, Roberto Garrappa and nicolettadelbuono! Confirmed speakers, among the others, F. Radu (PMG Bergen), R. Tempone (RWTH Aachen and #KAUST), M. Ferronato (Università di Padova). Don’t miss it! intrusion2023.irsa.cnr.it

I'm glad to share our new paper on Hydrological Processes. With some colleagues at CNR IRSA we modeled the performances of a cluster of interconnected infiltration basins, used as final receptor of the effluents from a tertiary treatment wastewater plant.

Finally online! Great job @lauraselicato nicolettadelbuono @rafalzdunek sciencedirect.com/science/articl…

#NewProfilePic

Una panoramica su Planetek Italia su La Repubblica Bari. Dai concept satellitari innovativi, ai servizi per la PA e l'ingegneria, l'IA applicata alle immagini satellitari, la nuova costellazione italiana IRIDE. PS: cercasi talenti per collaborare in queste importanti sfide 🧑‍🎓

✔️Finally published our paper on Elsevier Mathematics #CNSNS, with #GiovanniGirardi, previously post-doc CNR IRSA, now UnivPM. We mathematically frame the ecological memory of water stress by plant roots, modeling roots water absorption by a #nonlocal sink term.

#8AYW #Day2 And here's the official group picture of this edition 😍

#8AYW #Day3 #ApplicationsOfOR Flavia Esposito enhances a Singular Value Decomposition for Nonnegative Matrix Factorization leveraging the low-rank structure of the residual matrix, maintaining high-quality initializations while reducing computational complexity #orms

Here is my another poster for NOPTA in Belgium, joint work with Flavia Esposito TLDR - Instead of Euclidean distance, we switch to distance on sphere for NMF - using Riemannian gradient method - we propose a cheap projection-free update for H - still many open problems ...

Read if you care about NMF Chordal-NMF with Riemannian Multiplicative Update What - NMF is not linear but conic algebra - Point-wise distance not good for cone - We propose chrodal / angle distance - We propose a Riemannian gradient method for this NMF - Fun arXiv 2405.12823

Sum-of-norms NMF - esimtate nonnegative rank (empirically) on-the-fly - how: overstimate the rank, the algo will fuse the components - Optimization non-trivial: nonsmooth nonconvex non-separable non-proximal non-dualizable - some theory & applications arXiv draft: 2407.00706

Vacation Relaxation?

Finally online! Joint work with Nicolas Gillis and Syed Atif link.springer.com/article/10.100…

Our latest paper about #bilevel optimization for #hyperparameter tuning is online. Too happy to work again with @lauraSelicato nicolettadelbuono @rafalzdunek Read more on: sciencedirect.com/science/articl…

Colleague: what’s your plan for Christmas Me: spent all the time to turn these books into nightmare to torture some poor souls in Italy next month Colleague: amen.

Sparse Hyperparametric IS NMF via Bi-Level Optimization - NMF with regularization - nobody likes tuning parameter so auto-tune - auto-tune via bi-level opt with hypergrad & ODE - solve it let you detect broken gear in machine - no theory, too complicated arXiv 2502.17123

This is really exciting and impressive, and this stuff is in my area of mathematics research (convex optimization). I have a nuanced take. 🧵 (1/9)

Our paper published (lnkd.in/ecw_vaJW) We have an open problem - tried Dinkelbach fractional program with Jagannathan’s iteration, not good - tried manifold but only n=1 (useless) - tried consensus (like ADMM), not good I think this problem is fun but I don't have time

[Research] Together with 2 strong Italian mothers, we try to tackle structrual bi-level nonsmooth nonconvex problem. This problem finds application in machine learning I am not strong enough for this hard problem, results are preliminary arXiv avaliable next Monday (fix typo)