Scientific Calendar Event
GO TO SCIENTIFIC CALENDAR



  • Europe/Rome
  • English
  • Administration
Description An ICTP Hybrid Meeting

The seventh edition of “Youth in high dimensions” will give to young researchers the opportunity to gather and present recent results on high-dimensional statistical problems, in areas ranging from machine learning and inference to neuroscience or statistical physics.

Typical problems in high-dimensions include estimating noisy signals (compressed sensing, PCA and tensor decomposition), analysing (deep) neural-networks, detecting communities in large networks, or modelling disordered spin systems. These seemingly unrelated problems share many similarities in their phenomenology and in the tools used to analyse them. The core of the field is made of a very active, diverse and quickly expanding community of physicists, computer scientists, mathematicians, information theorists and engineers, with the common desire to tackle increasingly challenging problems at the forefront of data science.

This conference aims at reinforcing the links among this interdisciplinary community, and in particular of its youngest theory-oriented members, and to bring forward the latest development happening in the high-dimensional world.

Topics
  • High-dimensional statistics
  • Statistical physics
  • Machine learning
  • Computational neuroscience
  • Complex systems
  • Machine learning for science
  • Statistical inference
Speakers
P. Beneventano, MIT, USA
S. D. Buchanan, UC Berkeley, USA
J. Cohen, Flatiron, USA
R. Cosson, NYU, USA
Y. Dandi, EPFL, Switzerland
C. Domine, ISTA, Austria
A. Favero, Cambridge University, UK
J. Gaitonde, Duke University, USA
A. Koloskova, Zurich University, Switzerland
G. Kur, ETH, Switzerland
C. Lauditi, Harvard University, USA
S. Li, Yale University, USA
A. Lotfi, Apple, Switzerland
P. Marion, INRIA Paris, France
E. Nichani, Princeton University, USA
L. Noci, ETH, Switzerland
R. Skerk, ICTP, Italy
N. Tsilivis, NYU, USA

Research abstracts: Applicants are encouraged to submit an abstract for a poster presentation. ​During the application, please make sure to use our templates to format your abstract. Templates are available below for download.
 
Contributed talk: The possibility of giving a contributed talk will be offered to a selection of people who will present a poster.

📢 The students are invited to follow the ceremony for the AI prize, that will be hosted at ICTP on July 23rd.

📢 The following activity to be held from 23 to 31 July 2026 may also be of interest for you: Advanced School in Applied Machine Learning: Uncertainty and Robustness of DL Models for Science | (smr 4228)

Grants: A limited number of grants are available to support the attendance of selected participants, with priority given to participants from developing countries. There is no registration fee

Material

Youth in High-Dimensions: Recent Progress in Machine Learning, High-Dimensional Statistics and Inference | (smr 4227)
Go to day
  • Monday, 20 July 2026
    • 08:45 - 09:00
      • 08:45 Registration Formalities 15'
    • 09:00 - 17:30
      • 09:00 Opening Remarks 5'
        Speaker: Jean BARBIER (ICTP, Italy), Elisabetta CORNACCHIA (Bocconi University, Italy), Sebastian Dominik GOLDT (SISSA, Italy), Marco MONDELLI (Institute of Science and Technology, Austria)
      • 09:05 How Muon Improves Fact Storage: A Study Through Linear Associative Memory 40' 
        Chairperson: Sebastian GOLDT (Sissa, Italy)
        Speaker: Eshaan NICHANI (Princeton University, USA)
        Material: Video
      • 09:45 Edge Flow: A Tractable and Predictive Continuous-Time Model for Gradient Descent at the Edge of Stability 45'
        Speaker: Pierre MARION (INRIA, France)
        Material: Video
      • 10:30 Coffee Break 30' ( Leonardo Building Terrace )
      • 11:00 Optimization algorithms for distributed and federated learning 45' 
        Chairperson: Marco MONDELLI (Institute of Science and Technology, Austria)
        Speaker: Anastasia KOLOSKOVA (Zurich University, Switzerland)
        Material: Video
      • 11:45 Collective exploration, entropic regularization, and random games 45'
        Speaker: Romanin COSSON (New York University, USA)
        Material: Video
      • 12:30 Lunch Break 1h30'
      • 14:00 Minimum Norm Interpolation via the Local Theory of Banach Spaces: The Role of Gaussianity 45' 
        Chairperson: Elisabetta CORNACCHIA (Bocconi Uiversity, Italy)
        Speaker: Gil KUR (ETH, Zurich)
        Material: Video
      • 14:45 Stable Algorithms Cannot Reliably Find Isolated Perceptron Solutions 45'
        Speaker: Shuangping LI (Yale University, USA)
        Material: Video
      • 15:30 Coffee Break 30' ( Leonardo Building Terrace )
      • 16:00 Poster session 1h30'
  • Tuesday, 21 July 2026
    • 09:00 - 16:00
      • 09:00 What Makes a Good Chain of Thought, and How Can Models Learn It Efficiently 45'
        Speaker: Aryo LOTFI (Apple, Switzerland)
        Material: Video
      • 09:45 Learning through Internalization 45' 
        Chairperson: Jean BARBIER (ICTP, Italy)
        Speaker: Nikolaos TSILIVIS (New York University, USA)
        Material: Video
      • 10:30 Coffee Break 30' ( Leonardo Building Terrace )
      • 11:00 Contributed Talks 1h30' 
        Chairperson: Sebastian GOLDT (Sissa, Italy)
      • 11:01 Statistical inference with correlated disorder in community detection 17'
        Speaker: Andrea Vincenzo DELL'ABATE (École normale supérieure, France)
        Material: Abstract Video
      • 11:18 Diverse Vision Transformers align with humans and one another in representing textures 18'
        Speaker: Ludovica DE PAOLIS (SISSA, Italy)
        Material: Abstract Video
      • 11:36 Interacting perceptrons: a minimal model of spontaneous collective learning 18'
        Speaker: Jerome Mathieu Samuel GARNIER BRUN (Bocconi University, Italy)
        Material: Abstract Video
      • 11:54 Factual recall in linear associative memories: sharp asymptotics and mechanistic insights 18'
        Speaker: Alessio GIORLANDINO (SISSA, Italy)
        Material: Abstract Video
      • 12:12 Complex non-linear effects via AMP 18'
        Speaker: Kevin KOEGLER (Institute of Science and Technology (ISTA), Austria)
        Material: Abstract Video
      • 12:30 Lunch Break 1h30'
      • 14:00 Spectral Dynamics in Deep Networks: Feature Learning, Outlier Escape, and Learning Rate Transfer 45' 
        Chairperson: Marco MONDELLI (Institute of Science and Technology, Austria)
        Speaker: Clarissa LAUDITI (Harvard University, USA)
        Material: Video
      • 14:45 Transformer Networks as Unrolled Optimization 45'
        Speaker: Sam D. BUCHANAN (UC Berkeley, USA)
        Material: Video
      • 15:30 Coffee Break 30' ( Leonardo Building Terrace )
    • 18:00 - 20:00
      • 18:00 Welcome Reception 2h0' ( Pane Quotidiano, Barcola ) 
        All Participants are cordially invited to the Welcome Reception
  • Wednesday, 22 July 2026
    • 09:00 - 16:05
      • 09:00 Contributed Talks 1h30' 
        Chairperson: Marco MONDELLI (Institute of Science and Technology, Austria)
      • 09:01 Stochastic Lanczos Quadrature for Computational Uncertainty in Linear Algebra 18'
        Speaker: Fifen HASSAN (African Institute for Mathematical Sciences Research and Innovation Centre, Rwanda)
        Material: Video
      • 09:18 Analyzing Representation Geometry of Speech SSL Models for Hinglish Code-Switched Speech 18'
        Speaker: Ali Hussaini UMAR (ICTP, Italy)
        Material: Video
      • 09:36 Theory of flexible inference of Langevin dynamics from high-dimensional noisy observations 18'
        Speaker: Yuzheng LIN (Princeton University, USA)
        Material: Video
      • 09:54 Multitask learning in the proportional limit 18'
        Speaker: Theo MARCHETTA (University of Bologna, Italy)
        Material: Video
      • 10:12 Stab-SGD: Noise-adaptivity in Smooth Optimization with Stability Ratios 18'
        Speaker: David Albert Rolando ROBIN (Dauphine University, France)
        Material: Video
      • 10:30 Coffee Break 30' ( Leonardo Building Terrace )
      • 11:00 Statistical physics of deep learning: Optimal learning of a multi-layer perceptron near interpolation 45'
        Speaker: Rudy SKERK (ICTP, Italy)
        Material: Video
      • 11:45 Deep Learning as Neural Low-Degree Filtering: A Spectral Theory of Hierarchical Feature Learning 45'
        Speaker: Yatin DANDI (EPFL, Switzerland)
        Material: Video
      • 12:30 Lunch Break 1h30'
      • 14:00 Contributed Talks 1h30' 
        Chairperson: Jean BARBIER (ICTP, Italy)
      • 14:01 A Fourier perspective on the learning dynamics of neural networks: from sample complexities to mechanistic insights 17'
        Speaker: Fabiola RICCI (SISSA, Italy)
        Material: Video
      • 14:18 Emergence of Distortions in High-Dimensional Guided Diffusion Models 18'
        Speaker: Enrico VENTURA (SISSA, Italy)
        Material: Video
      • 14:36 Characterizing collective feature selectivity in populations of visual units 18'
        Speaker: Lorenzo TAUSANI (SISSA, Italy)
        Material: Video
      • 14:54 Generative Posterior Sampling for Inverse Problems in the Geosciences 18'
        Speaker: Chiara ZELCO (Ecole Normale Superieure, France)
        Material: Video
      • 15:12 Prompt-Dependent Phase Transitions in the Output Distribution of Large Language Models 18'
        Speaker: Nikolija CUCKIC (University of Niš, Serbia)
        Material: Video
      • 15:30 Group photo & Coffee break 30' ( Leonardo Building Terrace )
  • Thursday, 23 July 2026
    • 09:00 - 14:00
      • 09:00 A Theory of How Pretraining Shapes Inductive Bias in Fine-Tuning 45' 
        Chairperson: Elisabetta CORNACCHIA (Bocconi Uiversity, Italy)
        Speaker: Clementine DOMINE (ISTA, Austria)
        Material: Video
      • 09:45 Learn from your own latents, not from tokens: A sample-complexity theory 45'
        Speaker: Alessandro FAVERO (Cambridge University, UK)
        Material: Video
      • 10:30 Coffee Break 30' 
        Chairperson: Sebastian GOLDT (Sissa, Italy)
      • 11:00 How does gradient descent work? 45'
        Speaker: Jeremiah COHEN (FLATIRON INSTITUTE, USA)
        Material: Video
      • 11:45 How does SGD train neural networks? On the edge of (stochastic) stability 45' 
        Chairperson: Sebastian GOLDT (Sissa, Italy)
        Speaker: Pierfrancesco BENEVENTANO (Massachusetts Institute of Technology, USA)
        Material: Video
      • 12:30 Lunch Break 1h30'
    • 14:30 - 18:00 The ICTP-IBM Brahmagupta AI Prize for Early Career Scientists Ceremony 
      The Ceremony will be livestreamed at: ictp.it/livestream
      • 14:30 Opening remarks 20'
        Speaker: Atish Dabholkar (Director, ICTP, Italy), Alessandro CURIONI (IBM, Switzerland)
      • 14:50 Announcement and motivation by the jury 10' 
        Remote
        Speaker: Marc MEZARD (Bocconi University, Italy)
      • 15:00 Award ceremony 10'
      • 15:10 Laudatio 20' 
        Remote
        Speaker: David S. WISHART (University of Alberta, Canada)
      • 15:30 Charting the unknown metabolome with biochemical AI 45'
        Speaker: Michael SKINNIDER (Princeton University, USA)
        Material: Abstract
      • 16:15 Small, Specialised, Collaborative and Open-Weight: The Models Enabling Physical AI 45'
        Speaker: Eiman KANJO (NTU, UK)
      • 17:00 Group Photo 5'
      • 17:05 Refreshments 55' ( Leonardo building Terrace )
GO TO SCIENTIFIC CALENDAR