Matthieu Wyart
12 papers · 2018–2025 · 3 venues · across top CS/AI venues
Authors are matched by name, so researchers who share a name ("Yang Liu", "Wei Wang") appear as a single profile and their paper counts, venues and achievements are combined. How identity is handled.
Achievements
Jump to papers ↓+5 more ↓ Show less ↑
π£ Hot Topic Early Bird π Conference Polyglot Β· Bronze π Academic Marathon Β· Sprint π Interdisciplinary Bridge Β· Silver πΊοΈ Taxonomy Completionist (25)
π§
Keyword Pioneer
ποΈ
Keyword Collector
(58)
π
Century Club Β· Bronze
π₯
Unstoppable Β· Bronze
β
The Questioner
Venues
Top co-authors
Research topics
Machine Learning
(11)
Optimization & Theory Β· Machine Learning
(11)
Deep Learning
(8)
Core Methods
(6)
Models
(4)
Architectures
(3)
Learning Types
(3)
Optimization & Theory Β· Deep Learning
(3)
Interdisciplinary
(2)
Linguistics
(2)
Artificial Intelligence
(1)
Mathematics & Optimization
(1)
Natural Language Processing
(1)
Core AI
(1)
Generation
(1)
Keywords
generalization error
(4)
sample complexity
(2)
probabilistic context-free grammar
(2)
feature learning
(2)
convolutional neural network
(2)
hierarchical structure
(2)
diffusion model
(2)
learning curve
(2)
lazy training
(2)
neural network
(2)
sparse representation
(1)
compositional generalization
(1)
hierarchical representation
(1)
eigenvalue decomposition
(1)
loss landscape
(1)
grammar learning
(1)
image classification
(1)
neural network optimization
(1)
replica method
(1)
deep learning
(1)
Papers
How Compositional Generalization and Creativity Improve as Diffusion Models are Trained
ICML 2025
Learning curves theory for hierarchically compositional data with power-law distributed features
ICML 2025
Probing the Latent Hierarchical Structure of Data via Diffusion Models
ICLR 2025
How Deep Networks Learn Sparse and Hierarchical Data: the Sparse Random Hierarchy Model
ICML 2024
Towards a theory of how the structure of language is acquired by deep neural networks
NeurIPS 2024
Dissecting the Effects of SGD Noise in Distinct Regimes of Deep Learning
ICML 2023
What Can Be Learnt With Wide Convolutional Neural Networks?
ICML 2023
Failure and success of the spectral bias prediction for Laplace Kernel Ridge Regression: the case of low-dimensional data
ICML 2022
Learning sparse features can lead to overfitting in neural networks
NeurIPS 2022
Locality defeats the curse of dimensionality in convolutional teacher-student scenarios
NeurIPS 2021
Relative stability toward diffeomorphisms indicates performance in deep nets
NeurIPS 2021
Comparing Dynamics: Deep Neural Networks versus Glassy Systems
ICML 2018