Seminars

Frank–Wolfe and friends: a journey into projection-free optimization methods

Francesco Rinaldi
Università degli studi di Padova

Edge of Stochastic Stability: SGD does not train neural networks as you expect it

Pierfrancesco Beneventano
Massachusetts Institute of Technology (MIT)

Statistical Properties of Rectified Flow

Arun Kuchibhotla
Carnegie Mellon University

From Score Matching to Diffusion: a fine-grained error analysis in the Gaussian setting

Samuel Hurault
ENS Paris

Optimal Transport-based Conformal Prediction

Kimia Nadjahi
ENS Paris

Unsupervised Imaging Inverse Problems with Diffusion Distribution Matching

Giacomo Meanti
INRIA Grenoble

Bayesian Imaging in the Low-Photon Regime: Data-Driven Priors and Algorithms

Teresa Klatzer
University of Edinburgh

Large-time dynamics in transformer architectures with layer normalisation

Yury Korolev
University of Bath

A preconditioned second-order convex splitting algorithm with a difference of varying convex functions and line search

Hongpeng Sun
Renmin University of China

Nonparametric two-sample testing with permutations, kernels and optimal transport.

Nicolas Schreuder
CNRS

A simpler characterization of GBD

Antonin Chambolle
CNRS senior scientist at CEREMADE, Université Paris-Dauphine, Paris

Wasserstein Gradient Flow on the Maximum Mean Discrepancy

Arthur Gretton
University College London, Google DeepMind

On tomographic imaging with limited data

Tatiana Bubba
Università degli Studi di Ferrara

TFML Talk: A mean-field view on transformer models

Andrea Agazzi
Institute of Mathematical Statistics and Actuarial Science - Universität Bern

TFML Talk: The algorithmic foundations of online learning

Nicolò Cesa-Bianchi
Università degli Studi di Milano

TFML Talk: (Sparse) Compositionality

Tomaso Poggio
Massachusetts Institute of Technology

Self-supervised learning for inverse problems

Julian Tachella
ENS Lyon

Principled approaches and tools for the analysis and design of first-order optimization algorithms

Adrian Taylor
INRIA, École Normale Supérieure

TomoSelfDEQ: a Self-Supervised DEQ approach for Sparse-Angle CT

Andrea Sebastiani
University of Modena and Reggio Emilia - Unimore

Theoretical considerations for practical meta reinforcement learning

Mirco Mutti
Technion

MaLGa Colloquia: Cross-modal generation and understanding of multimodal content

Niculae Sebe
Università di Trento

Unfolded proximal neural networks for computational imaging

Audrey Repetti
Heriot Watt University, Edinburgh, UK

MaLGa Colloquia - Rapture of the deep: highs and lows of sparsity in a world of depths

Rémi Gribonval
Ecole Normale Supérieure, Lyon (France)

Geometric and computational complexity in bilevel optimization

Samuel Vaiter

Spectral complexity of deep neural networks

Stefano Vigogna
Università di Roma Tor Vergata

Hadamard Langevin dynamics for the l1 priors

Clarice Poon
University of Warwick

MaLGa Colloquia: The Emerging Science of Machine Learning Benchmarks

Moritz Hardt
Max Planck Institute for Intelligent Systems, Tübingen, Germany

MaLGa Colloquia: Science of Intelligence, a personal perspective

Tomaso Poggio
Massachusetts Institute of Technology

Sparse and Low-Dimensional Structures in Deep Networks

Akshay Rangamani
New Jersey Institute of Technology

Reproducing Kernels in and for the Mean Field Limit

Christian Fiedler
RWTH Aachen University

Fisher-Rao metric for Gaussian processes

Minh Ha Quang
RIKEN Center for Advanced Intelligence Project (RIKEN-AIP)

Operator World-models for Reinforcement Learning

Carlo Ciliberto
University College London

Insights into the Role of the Initialisation and Curriculum through Parity Targets

Elisabetta Cornacchia
INRIA Paris, Argo project-team

MaLGa Colloquia - Towards a synergistic human-machine interaction and collaboration: XAI and Hybrid Decision Making Systems. State-of-the-art and research questions.

Fosca Giannotti
Scuola Normale Superiore (SNS), Pisa, Italy

Deep learning theory through the lens of diagonal linear networks

Scott Pesme
Inria Grenoble

Resilient Artificial Intelligence in Medical Imaging: Handling and Mining Multiple Sources

Michela Gravina
University of Naples Federico II

Don’t be so Monotone: Relaxing Stochastic Line Search in Over-Parameterized Models

Leonardo Galli
LMU Munich

Generalization of Hamiltonian algorithms

Andreas Maurer

From Computer Vision to Artificial Intelligence

Algorithms, Data, Knowledge and Reasoning

Pietro Perona

From known knowns to unknown unknowns in AI: Historical and Technical Issues

Fabio Roli
University of Genoa

MaLGa Colloquia - Some non-parametric Results

Yoav Freund
University of California San Diego (UCSD)

Can You Hear the Shape of a Room?

Tom Sprunck
University of Strasbourg (France)

Unveiling convolutional neural networks in surrogate modeling (and more!)

Nicola Rares Franco
Politecnico di Milano, MOX laboratory

Statistical and computational guarantees for gradient based MCMC in some PDE inverse problems

Richard Nickl
University of Cambridge

Gaussian Approximation and Bayesian Posterior Distribution in Random Deep Neural Networks

Dario Trevisan
Unversity of Pisa

Time-frequency analysis and discrete representation of functions and distributions

Gianluca Giacchi
University of Bologna - University of Lausanne

Convex Formulation for Total-Variation Parameter Learning

Enis Chenchene

The role of depth in neural networks: function space geometry and learnability

Rebecca Willett
University of Chicago

Representation Equivalent Neural Operators: a Framework for Alias-free Operator Learning

Francesca Bartolucci
Delft Institute of Applied Mathematics (TU Delft)

Over-parameterization in (two-layer) neural networks: double descent, function spaces, curse of dimensionality

Fanghui Liu
University of Warwick

Accelerated optimization algorithms: the good, the bad and the odd

Juan Peypouquet
University of Groningen

AI4Science: Opportunities and Challenges

Luca Biggio
EPFL - Ecole polytechnique fédérale de Lausanne

Towards trustworthy digital patient monitoring: deep learning approaches for analyzing multimedia data from the actual clinical practice

Lucia Migliorelli
Università Politecnica delle Marche

Open World: Generalize and Recognize Novelty

Tatiana Tommasi
Politecnico di Torino

Dynamical low-rank training

Francesco Tudisco
University of Edinburgh, UK

Physics Priors for Machine Learning and Machine Learning to Solve Physics

Max Welling
University of Amsterdam

Sparsistency guarantees for inverse optimal transport

Clarice Poon
University of Warwick

ML for neuroscience and reproducibility for ML: from bilevel optimization to benchopt

Mathurin Massias
INRIA Lyon, Ockham Team

A rainbow in deep network black boxes

Florentin Guth
École Normale Supérieure, Paris

Sampling with Langevin Algorithms in Continuous and Discrete Times

Andre Wibisono
Yale University

Reproduction in a Toxic World

Jennifer Fung
University of California San Francisco, California

MaLGa Colloquia - The World of Graph Neural Networks: From the Mystery of Generalization to Foundational Limitations

Gitta Kutyniok
Ludwig-Maximilians Universität München

MaLGa Lecture - La matematica fra modelli fisici e intelligenza artificiale

Alfio Quarteroni
Politecnico di Milano

MaLGa Colloquia - The Statistical Complexity of Interactive Decision Making

Alexander (Sasha) Rakhlin
Massachusetts Institute of Technology

Revisiting optimization: gradient descent with a general cost

Pierre-Cyril Aubin-Frankowski
INRIA Paris (Sierra)

MaLGa Colloquia - Cancer Screening: A Parable of Prediction

Benjamin Recht
University of California, Berkeley

Convergence and optimality of wide RNNs in the mean-field regime

Andrea Agazzi
Università di Pisa

MaLGa Colloquia - Fast unrolled proximal algorithms to design stable and efficient neural network architectures

Nelly Pustelnik
CNRS, ENS de Lyon

MaLGa Colloquia - Modeling shapes and surfaces: geometry meets machine learning

Sayan Mukherjee
MPI Leipzig

What Neural Networks Memorize and Why

Vitaly Feldman
Apple ML Research

MaLGa Colloquia - Context-aware motion prediction and embodied social navigation

Lamberto Ballan
University of Padova

A numerical scheme to solve the multimarginal optimal transport with Coulomb cost

Rodrigue Lelotte
Université Paris-Dauphine

MaLGa Colloquia - Risk, Replay and Rehearsal (Remote)

Peter Dayan
MPI Tübingen

Regularized information geometric and optimal transport distances for Gaussian processes

Minh Ha Quang
RIKEN Center for Advanced Intelligence Project (AIP)

MaLGa Colloquia - The uses of memory in olfactory search

Antonio Celani
ICTP

Possible directions to deal with EEG data with machine learning

Andrea Apicella
Università di Napoli Federico II

Mortality containment vs. economics opening: optimal policies in a SEIARD model

Elena Beretta
New York University Abu Dhabi

Learning a microlocal prior for limited angle tomography

Tatiana Bubba
University of Bath

Graph and distributed extensions of the Douglas- Rachford method

Emanuele Naldi
TU Braunschweig

Model-based and data-driven approaches for modelling human sensory-motor system:

applications to advanced sensing systems and autonomous robotic grasping

Matteo Bianchi
Centro E. Piaggio, Università di Pisa

The implicit bias of gradient descent and its applications

Johannes Maly
Catholic University of Eichstaett-Ingolstadt

Stochastic Normalizing Flows for Inverse Problems: a Markov Chains Viewpoint

Johannes Hertrich
TU Berlin

Nonsmooth implicit differentiation for optimization

Edouard Pauwels
Toulouse 3 Paul Sabatier

The Radon transform, neural networks and splines

Michael Unser
EPFL

Results on the use of group equivariant non-expansive operators

for topological data analysis and geometric deep learning

Patrizio Frosini
Università di Bologna

Quiver representations and neural networks

Claudio Bartocci
Università di Genova

Learning-augmented count-min sketches via Bayesian nonparametrics

Stefano Favaro
Università di Torino - Collegio Carlo Alberto

Brain imaging with magneto/electro-encephalography

from source localization to connectivity estimation

Alberto Sorrentino
Università di Genova

Thoughts on today's learning theory - Tomaso Poggio #malgaseminar

Tomaso Poggio
MIT

Deep learning based reduced order models for numerical approximation of PDEs

Andrea Manzoni
Politecnico di Milano

Beyond Action Recognition: Detailed Video Modeling

Gül Varol
École des Ponts ParisTech

Sparsity and convergence analysis of generalized conditional gradient methods

Marcello Carioni
University of Cambridge

Three common reinforcement learning tricks: when and why do they work

He Niao
ETH Zurich

COVID-19: modelli e indicatori per provvedimenti di sanità pubblica

Stefano Merler
Fondazione Bruno Kessler

Deep Learning in Computational Imaging

Felix Lucka
Centrum Wiskunde & Informatica (Amsterdam)

Reinforcement Learning for Animal Behavior

Massimo Vergassola
CNRS

Non-Stationary Delayed Bandits with Intermediate Observations

Claire Vernade
DeepMind (UK)

Three Mathematical Tales of Machine Learning

Massimo Fornasier
Technical University of Munich

Computer Vision to Digitalise the Retail: Pipelines, Infrastructure & Open Problems.

Federico Roncallo
Trax Retail

Computational Cellular Engineering

Simone Bianco
IBM

Proximal and Invertible Neural Networks

Gabriele Steidl
TU Berlin

Critical Points, Multiple Testing and Point Source Detection for Cosmological Data

Domenico Marinucci
University of Rome Tor Vergata

On the use of 3D Gray Code Kernels for motion-related tasks in videos

Elena Nicora
DIBRIS, University of Genoa

Meaningful data and semantic interoperability: utopia or a possible reality (in the Italian Public Sector)?

Giorgia Lodi
STLab - Istituto di Scienze della Cognizione, CNR

Parameter-free Stochastic Optimization of Variationally Coherent Functions

Francesco Orabona
Boston University

Towards imitation learning in robotics

Luca Garello
DIBRIS, Università di Genova e Istituto Italiano di Tecnologia

Regularity Properties of Entropical Optimal Transport

Giulia Luise
Imperial College

Regularity properties of Entropic Optimal Transport in applications to machine learning

Giulia Luise
Imperial College

Foundations of deep convolutional models through kernel methods

Alberto Bietti
New York University

Towards Causal Representation Learning

Francesco Locatello
Amazon

Data Driven Regularization

Andrea Aspri
The Johann Radon Institute for Computational and Applied Mathematics

Self-supervised learning of depth and motion from monocular images

Junhwa Hur
Former Ph.D Student at TU-Darmstadt

Analysis of Gradient Descent on Wide Two-Layer ReLU Neural Networks

Lénaïc Chizat
Laboratoire de mathématiques d'Orsay at Université Paris-Saclay

Phase Retrieval of Bandlimited Functions for the Wavelet Transform

Francesca Bartolucci
ETH Zürich

Cognitive robotics for collaboration

Alessandra Sciutti
IIT

Dispersion, spreading and sparsity of Gabor wave packets

Ivan Salvatore Trapasso
DIMA, Università di Genova

Deep neural networks for inverse problems with pseudodifferential operators: an application to limited-angle tomography

Luca Ratti
University of Helsinki

Labelling actions in videos

Davide Moltisanti
Nanyang Technological University, Singapore

On the Happy Marriage of Kernel Methods and Deep Learning

Julien Mairal
INRIA Grenoble

Nonlinear Mean Value Properties and PDE's

Ángel Arroyo
Università di Genova

Machine Learning for Cellular Biosensing

Vito Paolo Pastore
Università di Genova

Learning the Invisible: Limited Angle Tomography, Shearlets and Deep Learning

Tatiana Bubba
University of Helsinki

3D scene and object understanding in the era of deep learning

Federico Tombari
Technische Universität München

Unitarization of the Radon transform on homogeneous trees

Matteo Monti
DIMA, Università di Genova

Manifold Structured Prediction: Theory and applications

Gian Maria Marconi
DIBRIS, Università di Genova

Markerless gait analysis in stroke survivors based on computer vision and deep learning: a pilot study

Matteo Moro
DIBRIS, Università di Genova

The Banach Gelfand Triple and Fourier Standard Spaces

Hans G. Feichtinger
Institute of Mathematics, University of Vienna

Duality for the Entropic Optimal Transport Problem and Applications

Simone di Marino
DIMA, Università di Genova

A Peek at the Landscape of Dictionary Learning

Karin Schnass
Universität Innsbruck

Towards Safe Reinforcement Learning

Andreas Krause
ETH Zürich

Multi-Modal Sensors for Human Behavior Monitoring

Paolo Napoletano
Universita' degli Studi di Milano-Bicocca

Monotonic Gaussian process flow

Carl Heinrik Ek
University of Bristol

Learning Interaction laws in particle- and agent-based systems

Mauro Maggioni
Johns Hopkins University

Uniform estimation of nonlinear statistics

Andreas Maurer

Adaptive backtracking and acceleration of a forward-backward algorithm for strongly convex optimisation: convergence results and imaging applications

Luca Calatroni
I3S Laboratory, CNRS, Sophia Antipolis, France

Optimal data approximation with group invariances

Davide Barbieri
Universidad Autónoma de Madrid

Curvelet frame and photoacoustic reconstruction

Marta Betcke
University College London

Statistical Machine Learning and Optimisation Challenges for Brain Imaging at a Millisecond Timescale

Alexandre Gramfort
INRIA Saclay Research Center and CEA Neurospin

Machine Learning for Image Processing

Dong Hye Ye
Marquette University

Stay positive! The importance of better models in stochastic optimization

John Duchi
Stanford University

The Principle of Least Cognitive Action

Marco Gori
University of Siena

Resolution of Sobolev wavefront set and sparse representation of singular integral operator using shearlets

Swaraj Paul
Indian Institute of Technology Indore

The relation between the Cahn-Hilliard equation and CMC surfaces

Matteo Rizzi
Facultad de Ciencias Físicas y Matemáticas, Universidad de Chile

Optimization in inverse problems via inertial iterative regularization

Guillaume Garrigos
Université de Paris

Learning with discrete MAP-inference models for stereo and motion

Thomas Pock
Graz University of Technology

Learning to Adapt: Digging Deeper into Domain Adaptation for Visual Recognition in Real-world and Dynamic Environments

Elisa Ricci
University of Trento and Fondazione Bruno Kessler, Italy

Designing non-parametric activation functions: recent advances

Simone Scardapane
Sapienza University of Rome

Electrical impedance tomography and Calderon's inverse problem: a review

Matteo Santacesaria
University of Genoa

Pair-matching and sequential learning of communities

Cristophe Giraud
Paris-Sud University

On the shape of hypersurfaces with almost constant mean curvature

Giulio Ciraolo
Palermo University

Multiscale decompositions in imaging and inverse problems

Luca Rondi
University of Milan

Optimal transport and gradient flows

Giuseppe Savarè
University of Pavia

Overview on speech processing: fundamentals, techniques and applications

Zied Mnasri
DIBRIS

The multi-agent approach to artificial general intelligence

Andrea Tacchetti
Deep Mind

On the Existence and on the Role of Wide Flat Minima in Deep Learning

Riccardo Zecchina
Bocconi University