4 papers
Training Flow Matching: The Role of Weighting and Parameterization
Anne Gagneux, Ségolène Martin, Rémi Gribonval +1
We study the training objectives of denoising-based generative models, with a particular focus on loss weighting and output parameterization, including noise-, clean image-, and ve…
Intrinsic training dynamics of deep neural networks
Sibylle Marcotte, Gabriel Peyré, Rémi Gribonval
A fundamental challenge in the theory of deep learning is to understand whether gradient-based training can promote parameters belonging to certain lower-dimensional structures (e.…
The Generation Phases of Flow Matching: a Denoising Perspective
Anne Gagneux, Ségolène Martin, Rémi Gribonval +1
Flow matching has achieved remarkable success, yet the factors influencing the quality of its generation process remain poorly understood. In this work, we adopt a denoising perspe…
Convexity in ReLU Neural Networks: beyond ICNNs?
Anne Gagneux, Mathurin Massias, Emmanuel Soubies +1
Convex functions and their gradients play a critical role in mathematical imaging, from proximal optimization to Optimal Transport. The successes of deep learning has led many to u…