Teodora Reu

Approximation bias during marginalization over nuisance parameters in likelihood-based variational inference, and its impact on Bayesian solutions

Sixtine Dromigny, Xuebin Zhao, Teodora Reu, Paula Koelemeijer et autres

SUMMARY Likelihood-based variational inference (VI) methods have recently gained traction in various fields of geophysics as they can dramatically reduce the computational cost of estimating Bayesian posterior probability distributions of target parameter values compared with traditional approaches such as Markov chain Monte …

gb (code pays fourni par la source)

Gassmann

Sixtine Dromigny, Xuebin Zhao, Teodora Reu, Paula Koelemeijer et autres

An example repository that demonstrates several Bayesian inference approaches for a toy geophysical inverse problem based on a Gassmann equation as forward model.

gb, cn (code pays fourni par la source)

Gradient Variance Reveals Failure Modes in Flow-Based Generative Models

Teodora Reu, Sixtine Dromigny, Michael M. Bronstein, Francisco Vargas

Rectified Flows learn ODE vector fields whose trajectories are straight between source and target distributions, enabling near one-step inference. We show that this straight-path objective conceals fundamental failure modes: under deterministic training, low gradient variance drives memorization of arbitrary training pairings, even …

gb, us (code pays fourni par la source)

Topological Message Passing for Higher - Order and Long - Range Interactions

Lorenzo Giusti, Teodora Reu, Francesco Ceccarelli, Cristian Bodnar et autres

Graph Neural Networks (GNNs) have demonstrated remarkable success in learning from graph-structured data. However, they face significant limitations in expressive power, struggling with long-range interactions and lacking a principled approach to modeling higher-order structures and group interactions. Cellular Isomorphism Networks (CINs) recently …

it, gb, us (code pays fourni par la source)

Metric Flow Matching for Smooth Interpolations on the Data Manifold

Kacper Kapuśniak, Peter Potaptchik, Teodora Reu, Leo Yu Zhang et autres

Matching objectives underpin the success of modern generative models and rely on constructing conditional paths that transform a source distribution into a target distribution. Despite being a fundamental building block, conditional paths have been designed principally under the assumption of Euclidean geometry, …

Metric Flow Matching for Smooth Interpolations on the Data Manifold

Kacper Kapuśniak, Peter Potaptchik, Teodora Reu, Leo Yu Zhang et autres

CIN++: Enhancing Topological Message Passing

Lorenzo Giusti, Teodora Reu, Francesco Ceccarelli, Cristian Bodnar et autres

Graph Neural Networks (GNNs) have demonstrated remarkable success in learning from graph-structured data. However, they face significant limitations in expressive power, struggling with long-range interactions and lacking a principled approach to modeling higher-order structures and group interactions. Cellular Isomorphism Networks (CINs) recently …

To smooth a cloud or to pin it down: Guarantees and Insights on Score Matching in Denoising Diffusion Models

Francisco Vargas, Teodora Reu, Anna Kerekes

Denoising diffusion models are a class of generative models which have recently achieved state-of-the-art results across many domains. Gradual noise is added to the data using a diffusion process, which transforms the data distribution into a Gaussian. Samples from the generative model …

Graph Neural Networks for Breast Cancer Data Integration

Teodora Reu

International initiatives such as METABRIC (Molecular Taxonomy of Breast Cancer International Consortium) have collected several multigenomic and clinical data sets to identify the undergoing molecular processes taking place throughout the evolution of various cancers. Numerous Machine Learning and statistical models have been …