GUYOMARD , Marie

The use of Artificial Intelligence for precision medicine is constantly rising. Machine Learning is increasingly employed to personalize patient care pathways, such as predicting pathologies or prescribing appropriate medical treatments. Algorithmic Decision Systems developed for this purpose take into account the specific clinical and...

This article focuses on approximating an interpretable neural network with kernel logistic regression. We introduce a new kernel that directly stems from the architecture of the neural network. The decision rule resulting from a logistic regression applied to this kernel is modeled as an additive decomposition of univariate functions and is...

This paper proposes an understandable neural network whose score function is modeled as an additive sum of univariate spline functions. It extends usual understandable models like generative additive models, spline-based models, and neural additive models. An approximation of this neural network by a kernel logistic regression provides...

This paper proposes a non-linear binary classification model. Although linear classification methods are very popular in the field of personalized medicine because of their interpretability, they have proven to be too restrictive. Doctors are convinced of the need to quantify threshold effects for better predictions. Nevertheless, non-linear...

This paper proposes an understandable neural network whose score function is modeled as an additive sum of univariate spline functions. It extends usual understandable models like generative additive models, spline-based models, and neural additive models. It is shown that this neural network can be approximated by a logistic regression whose...

Cet article s'intéresse à la classification binaire à l'aide d'une régression logistique non-linéaire. Les modèles linéaires, simples et interprétables, sont très appréciés dans le domaine médical mais leurs performances restent très limitées lorsque les données sont complexes. Nous proposons de remplacer la fonction linéaire de la régression...

ReLU neural networks suffer from a problem of explainability because they partition the input space into a lot of polyhedrons. This paper proposes a constrained neural network model that replaces polyhedrons by orthotopes: each hidden neuron processes only a single component of the input signal. When the number of hidden neurons is large, we...

International audience

Ce papier propose une nouvelle approche ajustant les réseaux de neurones convolutifs appliqués sur des jeux de données déséquilibrés dont les proportions par classes sont incertaines. La règle de décision constitutant la sortie du réseau de neurones est remplacée par le classifieur Minimax dont la particularité est de chercher à égaliser les...

International audience

This paper proposes a new approach for dealing with imbalanced classes and prior probability shifts in supervised classification tasks. Coupled with any feature space partitioning method, our criterion aims to compute an almost-Bayesian randomized equalizer classifier for which the maxima of the class-conditional risks are minimized. Our...

International audience