Image de couverture de livre

A Phase Transition in Diffusion Models Reveals the Hierarchical Nature of Data

Fuente: arXiv

Enregistré dans:

Détails bibliographiques
Auteurs principaux:	Sclocchi, Antonio, Favero, Alessandro, Wyart, Matthieu
Format:	Preprint
Publié:	2024
Sujets:	Machine Learning Disordered Systems and Neural Networks Computer Vision and Pattern Recognition
Accès en ligne:	Acceder al recurso
Tags:	Ajouter un tag Pas de tags, Soyez le premier à ajouter un tag!

Documents similaires

Probing the Latent Hierarchical Structure of Data via Diffusion Models
par: Sclocchi, Antonio, et autres
Publié: (2024)

Scaling Laws and Representation Learning in Simple Hierarchical Languages: Transformers vs. Convolutional Architectures
par: Cagnetta, Francesco, et autres
Publié: (2025)

On the different regimes of Stochastic Gradient Descent
par: Sclocchi, Antonio, et autres
Publié: (2023)

How Deep Networks Learn Sparse and Hierarchical Data: the Sparse Random Hierarchy Model
par: Tomasini, Umberto, et autres
Publié: (2024)

Towards a theory of how the structure of language is acquired by deep neural networks
par: Cagnetta, Francesco, et autres
Publié: (2024)

Hierarchical Associative Memory, Parallelized MLP-Mixer, and Symmetry Breaking
par: Karakida, Ryo, et autres
Publié: (2024)

Sampling Data with Chains of Forward-Backward Diffusion Steps
par: Kang, Hyunmo, et autres
Publié: (2026)

Deep networks learn to parse uniform-depth context-free languages from local statistics
par: Parley, Jack T., et autres
Publié: (2026)

Learning curves theory for hierarchically compositional data with power-law distributed features
par: Cagnetta, Francesco, et autres
Publié: (2025)

Memorization to Generalization: Emergence of Diffusion Models from Associative Memory
par: Pham, Bao, et autres
Publié: (2025)

On the Emergence of Linear Analogies in Word Embeddings
par: Korchinski, Daniel J., et autres
Publié: (2025)

iMixer: hierarchical Hopfield network implies an invertible, implicit and iterative MLP-Mixer
par: Ota, Toshihiro, et autres
Publié: (2023)

The Physics of Data and Tasks: Theories of Locality and Compositionality in Deep Learning
par: Favero, Alessandro
Publié: (2025)

Symmetry in language statistics shapes the geometry of model representations
par: Karkada, Dhruva, et autres
Publié: (2026)

Architecture-Aware Minimization (A$^2$M): How to Find Flat Minima in Neural Architecture Search
par: Gambella, Matteo, et autres
Publié: (2025)

Envisioning Future Deep Learning Theories: Some Basic Concepts and Characteristics
par: Su, Weijie J.
Publié: (2021)

Boosting Classification with Quantum-Inspired Augmentations
par: Tschöpe, Matthias, et autres
Publié: (2025)

Critical Phase Transition in Large Language Models
par: Nakaishi, Kai, et autres
Publié: (2024)

Neural Vector Tomography for Reconstructing a Magnetization Vector Field
par: Butbaia, Giorgi, et autres
Publié: (2024)

Solving Inverse Problems in Stochastic Self-Organizing Systems through Invariant Representations
par: Najarro, Elias, et autres
Publié: (2025)

Theory of Speciation Transitions in Diffusion Models with General Class Structure
par: Achilli, Beatrice, et autres
Publié: (2026)

Training Multi-Layer Binary Neural Networks With Local Binary Error Signals
par: Colombo, Luca, et autres
Publié: (2024)

Nature-Inspired Local Propagation
par: Betti, Alessandro, et autres
Publié: (2024)

Grokking as a First Order Phase Transition in Two Layer Networks
par: Rubin, Noa, et autres
Publié: (2023)

Geometric Entropy and Retrieval Phase Transitions in Continuous Thermal Dense Associative Memory
par: Petrova, Tatiana, et autres
Publié: (2026)

The Interplay of Data Structure and Imbalance in the Learning Dynamics of Diffusion Models
par: Nicoletti, Flavio, et autres
Publié: (2026)

A Generative Diffusion Model for Amorphous Materials
par: Yang, Kai, et autres
Publié: (2025)

Optimal Spectral Transitions in High-Dimensional Multi-Index Models
par: Defilippis, Leonardo, et autres
Publié: (2025)

Dynamical Regimes of Multimodal Diffusion Models
par: Albrychiewicz, Emil, et autres
Publié: (2026)

Emergence of Distortions in High-Dimensional Guided Diffusion Models
par: Ventura, Enrico, et autres
Publié: (2026)

Benefiting from Quantum? A Comparative Study of Q-Seg, Quantum-Inspired Techniques, and U-Net for Crack Segmentation
par: Srinivasan, Akshaya, et autres
Publié: (2024)

Can Geometric Quantum Machine Learning Lead to Advantage in Barcode Classification?
par: Umeano, Chukwudubem, et autres
Publié: (2024)

Why Diffusion Models Don't Memorize: The Role of Implicit Dynamical Regularization in Training
par: Bonnaire, Tony, et autres
Publié: (2025)

Deep Learning as Neural Low-Degree Filtering: A Spectral Theory of Hierarchical Feature Learning
par: Dandi, Yatin, et autres
Publié: (2026)

Grokking as the Transition from Lazy to Rich Training Dynamics
par: Kumar, Tanishq, et autres
Publié: (2023)

Exploring the Energy Landscape of RBMs: Reciprocal Space Insights into Bosons, Hierarchical Learning and Symmetry Breaking
par: Toledo-Marin, J. Quetzalcóatl, et autres
Publié: (2025)

Diffusion Operator Geometry of Feedforward Representations
par: Reddy, Kanishka
Publié: (2026)

Modeling Structured Data Learning with Restricted Boltzmann Machines in the Teacher-Student Setting
par: Thériault, Robin, et autres
Publié: (2024)

Controlled Langevin Dynamics for Sampling of Feedforward Neural Networks Trained with Minibatches
par: Zambon, Alessandro, et autres
Publié: (2026)

Short-range depinning in the presence of velocity-weakening
par: de Geus, Tom W. J., et autres
Publié: (2024)