1st year PhD student in Machine Learning and Optimization.
Welcome to my personal website !
I am a first year PhD student, under the supervision of Axel Parmentier (CERMICS, Ecole des Ponts) and Mathieu Blondel (Google DeepMind). My research currently focuses on the integration of inexact oracles, such as combinatorial optimization heuristics, as layers in deep learning models.
Before that, I graduated from Ecole polytechnique (X2020) and got a master’s degree from Sorbonne Université in Learning and Algorithms (Master M2A).
Optimization layers in deep neural networks have enjoyed a growing popularity in structured learning, improving the state of the art on a variety of applications. Yet, these pipelines lack interpretability since they are made of two opaque layers: a highly non-linear prediction model, such as a deep neural network, and an optimization layer, which is typically a complex black-box solver. Our goal is to improve the transparency of such methods by providing counterfactual explanations. We build upon variational autoencoders a principled way of obtaining counterfactuals: working in the latent space leads to a natural notion of plausibility of explanations. We finally introduce a variant of the classic loss for VAE training that improves their performance in our specific structured context. These provide the foundations of CF-OPT, a first-order optimization algorithm that can find counterfactual explanations for a broad class of structured learning architectures. Our numerical results show that both close and plausible explanations can be obtained for problems from the recent literature.
@article{vivier--ardisson_cf-opt_2024,title={{CF}-{OPT}: {Counterfactual} {Explanations} for {Structured} {Prediction}},shorttitle={{CF}-{OPT}},url={http://arxiv.org/abs/2405.18293},urldate={2024-05-29},publisher={arXiv},author={Vivier-Ardisson, Germain and Forel, Alexandre and Parmentier, Axel and Vidal, Thibaut},month=may,year={2024},note={arXiv:2405.18293 [cs]},keywords={Computer Science - Machine Learning},}
Learning with Local Search MCMC Layers
Germain Vivier-Ardisson, Mathieu Blondel , and Axel Parmentier
Integrating combinatorial optimization layers into neural networks has recently attracted significant research interest. However, many existing approaches lack theoretical guarantees or fail to perform adequately when relying on inexact solvers. This is a critical limitation, as many operations research problems are NP-hard, often necessitating the use of neighborhood-based local search heuristics. These heuristics iteratively generate and evaluate candidate solutions based on an acceptance rule. In this paper, we introduce a theoretically-principled approach for learning with such inexact combinatorial solvers. Inspired by the connection between simulated annealing and Metropolis-Hastings, we propose to transform problem-specific neighborhood systems used in local search heuristics into proposal distributions, implementing MCMC on the combinatorial space of feasible solutions. This allows us to construct differentiable combinatorial layers and associated loss functions. Replacing an exact solver by a local search strongly reduces the computational burden of learning on many applications. We demonstrate our approach on a large-scale dynamic vehicle routing problem with time windows.
@article{vivier2025learning,title={Learning with Local Search MCMC Layers},author={Vivier-Ardisson, Germain and Blondel, Mathieu and Parmentier, Axel},journal={arXiv preprint arXiv:2505.14240},urldate={2025-05-21},publisher={arXiv},month=may,year={2025},note={arXiv:2505.14240 [cs]},keywords={Computer Science - Machine Learning},}