Francesco Cagnetta

20 papers A* 6Journal 14

Year	Rank	Type	Title / Venue / Authors
2026	J	`jnl`	Deep networks learn to parse uniform-depth context-free languages from local statistics. CoRR Jack T. Parley, Francesco Cagnetta, Matthieu Wyart
2026	J	`jnl`	Deriving Neural Scaling Laws from the statistics of natural language. CoRR Francesco Cagnetta, Allan Raventós, Surya Ganguli, Matthieu Wyart
2025	A*	`conf`	How Compositional Generalization and Creativity Improve as Diffusion Models are Trained. ICML Alessandro Favero, Antonio Sclocchi, Francesco Cagnetta, Pascal Frossard, Matthieu Wyart
2025	J	`jnl`	How compositional generalization and creativity improve as diffusion models are trained. CoRR Alessandro Favero, Antonio Sclocchi, Francesco Cagnetta, Pascal Frossard, Matthieu Wyart
2025	A*	`conf`	Learning curves theory for hierarchically compositional data with power-law distributed features. ICML Francesco Cagnetta, Hyunmo Kang, Matthieu Wyart
2025	J	`jnl`	Learning curves theory for hierarchically compositional data with power-law distributed features. CoRR Francesco Cagnetta, Hyunmo Kang, Matthieu Wyart
2025	J	`jnl`	Scaling Laws and Representation Learning in Simple Hierarchical Languages: Transformers vs. Convolutional Architectures. CoRR Francesco Cagnetta, Alessandro Favero, Antonio Sclocchi, Matthieu Wyart
2024	A*	`conf`	Towards a theory of how the structure of language is acquired by deep neural networks. NeurIPS Francesco Cagnetta, Matthieu Wyart
2024	J	`jnl`	Towards a theory of how the structure of language is acquired by deep neural networks. CoRR Francesco Cagnetta, Matthieu Wyart
2023	J	`jnl`	How Deep Neural Networks Learn Compositional Data: The Random Hierarchy Model. CoRR Leonardo Petrini, Francesco Cagnetta, Umberto M. Tomasini, Alessandro Favero, Matthieu Wyart
2023	J	`jnl`	How deep convolutional neural networks lose spatial information with training. Mach. Learn. Sci. Technol. Umberto M. Tomasini, Leonardo Petrini, Francesco Cagnetta, Matthieu Wyart
2023	J	`jnl`	Kernels, Data & Physics. CoRR Francesco Cagnetta, Deborah Oliveira, Mahalakshmi Sabanayagam, Nikolaos Tsilivis, Julia Kempe
2023	A*	`conf`	What Can Be Learnt With Wide Convolutional Neural Networks? ICML Francesco Cagnetta, Alessandro Favero, Matthieu Wyart
2022	J	`jnl`	How Wide Convolutional Neural Networks Learn Hierarchical Tasks. CoRR Francesco Cagnetta, Alessandro Favero, Matthieu Wyart
2022	J	`jnl`	How deep convolutional neural networks lose spatial information with training. CoRR Umberto M. Tomasini, Leonardo Petrini, Francesco Cagnetta, Matthieu Wyart
2022	A*	`conf`	Learning sparse features can lead to overfitting in neural networks. NeurIPS Leonardo Petrini, Francesco Cagnetta, Eric Vanden-Eijnden, Matthieu Wyart
2022	J	`jnl`	Learning sparse features can lead to overfitting in neural networks. CoRR Leonardo Petrini, Francesco Cagnetta, Eric Vanden-Eijnden, Matthieu Wyart
2021	A*	`conf`	Locality defeats the curse of dimensionality in convolutional teacher-student scenarios. NeurIPS Alessandro Favero, Francesco Cagnetta, Matthieu Wyart
2021	J	`jnl`	Locality defeats the curse of dimensionality in convolutional teacher-student scenarios. CoRR Alessandro Favero, Francesco Cagnetta, Matthieu Wyart
2021	J	`jnl`	PhyDSL Multim. Tools Appl. Maria Teresa Baldassarre, Danilo Caivano, Simone Romano, Francesco Cagnetta, Víctor Fernández-Cervantes, Eleni Stroulia