Joseph de Vilmarest
Research Subjects
Joseph de Vilmarest has a total of 16 co-authors. Here are the top co-authors by number of shared publications:
- Olivier Wintenberger
Sorbonne University Meteo Meteo France
9 shared publications - Yannig Goude
OSIRIS EDF R&D Bd Gaspard Monge 91120 Saclay France
3 shared publications - Jean-Luc Mahoromeza
Sorbonne Universite
2 shared publications - Jethro Browell
School of Mathematics and Statistics University of Glasgow Glasgow UK
2 shared publications - Adeline Fermanian
LOPF, Califrais’ Machine Learning Lab
1 shared publication - Adeline Fermanian
LabCom Large-scale Optimization of Product Flows (LOPF)
1 shared publication
Latest Publications
- 7 April 2026
- Published by Agence Bibliographique de l'Enseignement Supérieur
Modèles espace-état pour la prévision de séries temporelles. Application aux marchés électriques
L'électricité étant difficile à stocker, prévoir la demande est un enjeu majeur pour maintenir l'équilibre entre la production et la consommation. L'évolution des usages de l'électricité, le déploiement des énergies renouvelables, et plus récemment la crise du coronavirus, motivent l'étude de modèles qui évoluent au cours du temps, pour tenir compte des changements de comportements. L'objectif de ce travail est de proposer des méthodes adaptatives de prévision, et nous nous sommes intéressés tout spécialement au cadre des modèles espace-état. Dans ce paradigme, on représente l'environnement (ou le contexte) par un état caché. À chaque instant, la demande dépend de cet état que nous cherchons donc à estimer grâce aux observations dont nous disposons, et selon les hypothèses que l'on effectue sur la dynamique du système. L'estimation de l'état nous permet ensuite de prévoir la demande. Un premier objectif de la thèse est de contribuer au lien entre l'optimisation et l'estimation dans les modèles espace-état. Nous interprétons en effet les méthodes que nous utilisons comme diverses façons de paramétrer un algorithme de descente de gradient de second ordre, et nous avons détaillé ce lien dans un cas particulier. Une seconde contribution de la thèse est de proposer différentes méthodes d'estimation dans les modèles espace-état. Le principal enjeu nous semble être de définir la dynamique avec lequel évolue l'état, et nous proposons deux méthodes dans ce but. Le troisième apport de ce manuscrit est d'appliquer ces méthodes espace-état à la prévision de consommation d'électricité. Nos prévisions s'appuient sur des modèles de prévision existants, par exemple le modèle additif généralisé, que nous cherchons à adapter. Ainsi, nous tirons parti de certaines dépendances complexes capturées par les modèles existants, par exemple la sensibilité de la consommation d'électricité à la température, tout en profitant de la faculté d'adaptation des modèles espace-état.
- 7 April 2026
- Published by Agence Bibliographique de l'Enseignement Supérieur
Modèles espace-état pour la prévision de séries temporelles. Application aux marchés électriques
L'électricité étant difficile à stocker, prévoir la demande est un enjeu majeur pour maintenir l'équilibre entre la production et la consommation. L'évolution des usages de l'électricité, le déploiement des énergies renouvelables, et plus récemment la crise du coronavirus, motivent l'étude de modèles qui évoluent au cours du temps, pour tenir compte des changements de comportements. L'objectif de ce travail est de proposer des méthodes adaptatives de prévision, et nous nous sommes intéressés tout spécialement au cadre des modèles espace-état. Dans ce paradigme, on représente l'environnement (ou le contexte) par un état caché. À chaque instant, la demande dépend de cet état que nous cherchons donc à estimer grâce aux observations dont nous disposons, et selon les hypothèses que l'on effectue sur la dynamique du système. L'estimation de l'état nous permet ensuite de prévoir la demande. Un premier objectif de la thèse est de contribuer au lien entre l'optimisation et l'estimation dans les modèles espace-état. Nous interprétons en effet les méthodes que nous utilisons comme diverses façons de paramétrer un algorithme de descente de gradient de second ordre, et nous avons détaillé ce lien dans un cas particulier. Une seconde contribution de la thèse est de proposer différentes méthodes d'estimation dans les modèles espace-état. Le principal enjeu nous semble être de définir la dynamique avec lequel évolue l'état, et nous proposons deux méthodes dans ce but. Le troisième apport de ce manuscrit est d'appliquer ces méthodes espace-état à la prévision de consommation d'électricité. Nos prévisions s'appuient sur des modèles de prévision existants, par exemple le modèle additif généralisé, que nous cherchons à adapter. Ainsi, nous tirons parti de certaines dépendances complexes capturées par les modèles existants, par exemple la sensibilité de la consommation d'électricité à la température, tout en profitant de la faculté d'adaptation des modèles espace-état.
- 1 January 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Signal Processing
- Vol. 74, 3929-3941
Robust Estimation for Linear State-Space Models With Stochastic Covariates
- 1 January 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Signal Processing
- Vol. 74, 3929-3941
Robust Estimation for Linear State-Space Models With Stochastic Covariates
- Open Access
Assessing Extrapolation of Peaks Over Thresholds with Martingale Testing
We present the winning strategy for the EVA2025 Data Challenge, which aimed to estimate the probability of extreme precipitation events. These events occurred at most once in the dataset making the challenge fundamentally one of extrapolating extreme values. Given the scarcity of extreme events, we argue that a simple, robust modeling approach is essential. We adopt univariate models instead of multivariate ones and model Peaks Over Thresholds using Extreme Value Theory. Specifically, we fit an exponential distribution to model exceedances of the target variable above a high quantile (after seasonal adjustment). The novelty of our approach lies in using martingale testing to evaluate the extrapolation power of the procedure and to agnostically select the level of the high quantile. While this method has several limitations, we believe that framing extrapolation as a game opens the door to other agnostic approaches in Extreme Value Analysis.
Open AccessAssessing Extrapolation of Peaks Over Thresholds with Martingale Testing
We present the winning strategy for the EVA2025 Data Challenge, which aimed to estimate the probability of extreme precipitation events. These events occurred at most once in the dataset making the challenge fundamentally one of extrapolating extreme values. Given the scarcity of extreme events, we argue that a simple, robust modeling approach is essential. We adopt univariate models instead of multivariate ones and model Peaks Over Thresholds using Extreme Value Theory. Specifically, we fit an exponential distribution to model exceedances of the target variable above a high quantile (after seasonal adjustment). The novelty of our approach lies in using martingale testing to evaluate the extrapolation power of the procedure and to agnostically select the level of the high quantile. While this method has several limitations, we believe that framing extrapolation as a game opens the door to other agnostic approaches in Extreme Value Analysis.
- Open Access
Assessing Extrapolation of Peaks Over Thresholds with Martingale Testing
We present the winning strategy for the EVA2025 Data Challenge, which aimed to estimate the probability of extreme precipitation events. These events occurred at most once in the dataset making the challenge fundamentally one of extrapolating extreme values. Given the scarcity of extreme events, we argue that a simple, robust modeling approach is essential. We adopt univariate models instead of multivariate ones and model Peaks Over Thresholds using Extreme Value Theory. Specifically, we fit an exponential distribution to model exceedances of the target variable above a high quantile (after seasonal adjustment). The novelty of our approach lies in using martingale testing to evaluate the extrapolation power of the procedure and to agnostically select the level of the high quantile. While this method has several limitations, we believe that framing extrapolation as a game opens the door to other agnostic approaches in Extreme Value Analysis.
Open AccessAssessing Extrapolation of Peaks Over Thresholds with Martingale Testing
We present the winning strategy for the EVA2025 Data Challenge, which aimed to estimate the probability of extreme precipitation events. These events occurred at most once in the dataset making the challenge fundamentally one of extrapolating extreme values. Given the scarcity of extreme events, we argue that a simple, robust modeling approach is essential. We adopt univariate models instead of multivariate ones and model Peaks Over Thresholds using Extreme Value Theory. Specifically, we fit an exponential distribution to model exceedances of the target variable above a high quantile (after seasonal adjustment). The novelty of our approach lies in using martingale testing to evaluate the extrapolation power of the procedure and to agnostically select the level of the high quantile. While this method has several limitations, we believe that framing extrapolation as a game opens the door to other agnostic approaches in Extreme Value Analysis.
- Open Access
Robust Estimation for Linear State-Space Models with Stochastic Covariates
The Kalman-Bucy filter (KF) is used to estimate and predict state-space models. It frequently encounters uncertain covariates, such as those derived from noisy measurements of physical signals (e.g., sensor readings, weather data, etc.). Our aim is to account for their uncertainties and develop robust KF approaches when incorporating the uncertainties of the covariates into the model—that is, by treating them as stochastic variables with known distributions. To enhance the robustness of the KF viewed as a Bayesian method that estimates the posterior distribution of the system state conditioned on past observations, we propose two methods. The linear Bayes estimator, which yields an explicit, unbiased forecast with minimal variance. An optimal risk estimator, which is non-linear and remains robust even when the covariates are highly noisy. This second method is approximated using a sequential Monte Carlo technique.
Open AccessRobust Estimation for Linear State-Space Models with Stochastic Covariates
The Kalman-Bucy filter (KF) is used to estimate and predict state-space models. It frequently encounters uncertain covariates, such as those derived from noisy measurements of physical signals (e.g., sensor readings, weather data, etc.). Our aim is to account for their uncertainties and develop robust KF approaches when incorporating the uncertainties of the covariates into the model—that is, by treating them as stochastic variables with known distributions. To enhance the robustness of the KF viewed as a Bayesian method that estimates the posterior distribution of the system state conditioned on past observations, we propose two methods. The linear Bayes estimator, which yields an explicit, unbiased forecast with minimal variance. An optimal risk estimator, which is non-linear and remains robust even when the covariates are highly noisy. This second method is approximated using a sequential Monte Carlo technique.
- 3 May 2025
- Published by Springer Nature in Statistical Papers
- Vol. 66 (4) , 1-44
Online convex optimization for survival analysis: an adaptive and stochastic approach
- 3 May 2025
- Published by Springer Nature in Statistical Papers
- Vol. 66 (4) , 1-44
Online convex optimization for survival analysis: an adaptive and stochastic approach
- Open Access
- 22 June 2024
- Published by Elsevier in International Journal of Forecasting
- Vol. 41 (4) , 1514-1520
An adaptive volatility method for probabilistic forecasting and its application to the M6 financial forecasting competition
- 22 June 2024
- Published by Elsevier in International Journal of Forecasting
- Vol. 41 (4) , 1514-1520
Open AccessAn adaptive volatility method for probabilistic forecasting and its application to the M6 financial forecasting competition
- 30 May 2024
- Published by Springer Nature in Statistical Inference for Stochastic Processes
- Vol. 27 (3) , 839-860
Viking: variational Bayesian variance tracking
- 30 May 2024
- Published by Springer Nature in Statistical Inference for Stochastic Processes
- Vol. 27 (3) , 839-860
Viking: variational Bayesian variance tracking
- Open Access
Adaptive time series forecasting with markovian variance switching
Adaptive time series forecasting is essential for prediction under regime changes. Several classical methods assume linear Gaussian state space model (LGSSM) with variances constant in time. However, there are many real-world processes that cannot be captured by such models. We consider a state-space model with Markov switching variances. Such dynamical systems are usually intractable because of their computational complexity increasing exponentially with time; Variational Bayes (VB) techniques have been applied to this problem. In this paper, we propose a new way of estimating variances based on online learning theory; we adapt expert aggregation methods to learn the variances over time. We apply the proposed method to synthetic data and to the problem of electricity load forecasting. We show that this method is robust to misspecification and outperforms traditional expert aggregation.
Open AccessAdaptive time series forecasting with markovian variance switching
Adaptive time series forecasting is essential for prediction under regime changes. Several classical methods assume linear Gaussian state space model (LGSSM) with variances constant in time. However, there are many real-world processes that cannot be captured by such models. We consider a state-space model with Markov switching variances. Such dynamical systems are usually intractable because of their computational complexity increasing exponentially with time; Variational Bayes (VB) techniques have been applied to this problem. In this paper, we propose a new way of estimating variances based on online learning theory; we adapt expert aggregation methods to learn the variances over time. We apply the proposed method to synthetic data and to the problem of electricity load forecasting. We show that this method is robust to misspecification and outperforms traditional expert aggregation.
- 30 August 2023
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Power Systems
- Vol. 39 (2) , 4154-4163
Adaptive Probabilistic Forecasting of Electricity (Net-)Load
Electricityload forecasting is a necessary capability for power system operators and electricity market participants. Both demand and supply characteristics evolve over time. On the demand side, unexpected events as well as longer-term changes in consumption habits affect demand patterns. On the production side, the increasing penetration of intermittent power generation significantly changes the forecasting needs. We address this challenge in two ways. First, our setting is adaptive ; our models take into account the most recent observations available to automatically respond to changes in the underlying process. Second, we consider probabilistic rather than point forecasting; indeed, uncertainty quantification is required to operate electricity systems efficiently and reliably. Our methodology relies on the Kalman filter, previously used successfully for adaptive point load forecasting. The probabilistic forecasts are obtained by quantile regressions on the residuals of the point forecasting model. We achieve adaptive quantile regressions using the online gradient descent; we avoid the choice of the gradient step size considering multiple learning rates and aggregation of experts. We apply the method to two data sets: the regional net-load in Great Britain and the demand of seven large cities in the United States. Adaptive procedures improve forecast performance substantially in both use cases for both point and probabilistic forecasting.
- 30 August 2023
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Power Systems
- Vol. 39 (2) , 4154-4163
Adaptive Probabilistic Forecasting of Electricity (Net-)Load
Electricityload forecasting is a necessary capability for power system operators and electricity market participants. Both demand and supply characteristics evolve over time. On the demand side, unexpected events as well as longer-term changes in consumption habits affect demand patterns. On the production side, the increasing penetration of intermittent power generation significantly changes the forecasting needs. We address this challenge in two ways. First, our setting is adaptive ; our models take into account the most recent observations available to automatically respond to changes in the underlying process. Second, we consider probabilistic rather than point forecasting; indeed, uncertainty quantification is required to operate electricity systems efficiently and reliably. Our methodology relies on the Kalman filter, previously used successfully for adaptive point load forecasting. The probabilistic forecasts are obtained by quantile regressions on the residuals of the point forecasting model. We achieve adaptive quantile regressions using the online gradient descent; we avoid the choice of the gradient step size considering multiple learning rates and aggregation of experts. We apply the method to two data sets: the regional net-load in Great Britain and the demand of seven large cities in the United States. Adaptive procedures improve forecast performance substantially in both use cases for both point and probabilistic forecasting.