olivier sorba | Orsay - Academia.edu (original) (raw)
Papers by olivier sorba
HAL (Le Centre pour la Communication Scientifique Directe), May 31, 2021
The aim of the present study is to detect abrupt trend changes in the mean of a multidimensional ... more The aim of the present study is to detect abrupt trend changes in the mean of a multidimensional sequential signal. Directly inspired by papers of Fernhead and Liu ([4] and [5]), this work describes the signal in a hierarchical manner : the change dates of a time segmentation process trigger the renewal of a piece-wise constant emission law. Bayesian posterior information on the change dates and emission parameters is obtained. These estimations can be revised online, i.e. as new data arrive. This paper proposes explicit formulations corresponding to various emission laws, as well as a generalization to the case where only partially observed data are available. Practical applications include the returns of partially observed multi-asset investment strategies, when only scant prior knowledge of the movers of the returns is at hand, limited to some statistical assumptions. This situation is different from the study of trend changes in the returns of individual assets, where fundamental exogenous information (news, earnings announcements, controversies, etc.) can be used.
Pénalités minimales pour la sélection de modèle Mots Clefs : moindres carrés pénalisés, sélection... more Pénalités minimales pour la sélection de modèle Mots Clefs : moindres carrés pénalisés, sélection de modèle, pénalité minimale, segmentation de signal gaussien, estimation de densité, contraste pénalisé, détection de ruptures multiples, CART, arbres de régression Résumé : Dans le cadre de la sélection de modèle par contraste pénalisé, L. Birgé et P. Massart ont prouvé que le phénomène de pénalité minimale se produit pour la sélection libre parmi des variables gaussiennes indépendantes. Nous étendons certains de leurs résultats à la partition d'un signal gaussien lorsque la famille de partitions envisagée est susamment riche, notamment dans le cas des arbres de régression. Nous montrons que le même phénomène se produit dans le cadre de l'estimation de densité. La richesse de la famille de modèles est liée à une forme d'isotropie. De ce point de vue le phénomène de pénalité minimale est intrinsèque. Pour corroborer et illustrer ce point de vue, nous montrons que le même phénomène se produit pour une famille de modèles d'orientation aléatoire uniforme.
The aim of the present study is to detect abrupt trend changes in the mean of a multidimensional ... more The aim of the present study is to detect abrupt trend changes in the mean of a multidimensional sequential signal. Directly inspired by papers of Fernhead and Liu ([4] and [5]), this work describes the signal in a hierarchical manner : the change dates of a time segmentation process trigger the renewal of a piece-wise constant emission law. Bayesian posterior information on the change dates and emission parameters is obtained. These estimations can be revised online, i.e. as new data arrive. This paper proposes explicit formulations corresponding to various emission laws, as well as a generalization to the case where only partially observed data are available. Practical applications include the returns of partially observed multi-asset investment strategies, when only scant prior knowledge of the movers of the returns is at hand, limited to some statistical assumptions. This situation is different from the study of trend changes in the returns of individual assets, where fundamenta...
Dans le cadre de la selection de modele par contraste penalise, L. Birge and P. Massart ont prouv... more Dans le cadre de la selection de modele par contraste penalise, L. Birge and P. Massart ont prouve que le phenomene de penalite minimale se produit pour la selection libre parmi des variables gaussiennes independantes. Nous etendons certains de leurs resultats a la partition d'un signal gaussien lorsque la famille de partitions envisagees est suffisamment riche, notamment dans le cas des arbres de regression. Nous montrons que le meme phenomene se produit dans le cadre de l'estimation de densite. La richesse de la famille de modele s'apparente a une forme d'isotropie. De ce point de vue le phenomene de penalite minimale est intrinseque. Pour corroborer et illustrer ce point de vue, nous montrons que le meme phenomene se produit pour une famille de modeles d'orientation aleatoire uniforme.
regression tree in N (∼phylogenetic) [ (d− 1) (√ d+ √ d− 1 )2]D Lem. B.3 free line segmentation [... more regression tree in N (∼phylogenetic) [ (d− 1) (√ d+ √ d− 1 )2]D Lem. B.3 free line segmentation [ e n−1 D−1 ]D−1 1 Sec. B.2 regular binary in N 8 2 log(2) ≤ 1.39 Lem. B.4 free binary in N [( 3 + 2 √ 2 ) e n D ]D 2 Lem. B.6 regular quad in N [ 4 4 3 3 ]D 5 3 log(2)− log(3) ≤ 0.057 Lem. B.5 free quad in N [ 4e n D ]D 2 Lem. B.7 Table 16 Complexity of some types of recursive segmentation trees B.4 Enumerating the models 221
ArXiv, 2021
The aim of the present study is to detect abrupt trend changes in the mean of a multidimensional ... more The aim of the present study is to detect abrupt trend changes in the mean of a multidimensional sequential signal. Directly inspired by papers of Fernhead and Liu ([4] and [5]), this work describes the signal in a hierarchical manner : the change dates of a time segmentation process trigger the renewal of a piece-wise constant emission law. Bayesian posterior information on the change dates and emission parameters is obtained. These estimations can be revised online, i.e. as new data arrive. This paper proposes explicit formulations corresponding to various emission laws, as well as a generalization to the case where only partially observed data are available. Practical applications include the returns of partially observed multi-asset investment strategies, when only scant prior knowledge of the movers of the returns is at hand, limited to some statistical assumptions. This situation is different from the study of trend changes in the returns of individual assets, where fundamenta...
The aim of the present study is to detect abrupt trend changes in the mean of a multidimensional ... more The aim of the present study is to detect abrupt trend changes in the mean of a multidimensional sequential signal. Directly inspired by papers of Fernhead and Liu ([4] and [5]), this work describes the signal in a hierarchical manner : the change dates of a time segmentation process trigger the renewal of a piece-wise constant emission law. Bayesian posterior information on the change dates and emission parameters is obtained. These estimations can be revised online, i.e. as new data arrive. This paper proposes explicit formulations corresponding to various emission laws, as well as a generalization to the case where only partially observed data are available. Practical applications include the returns of partially observed multi-asset investment strategies, when only scant prior knowledge of the movers of the returns is at hand, limited to some statistical assumptions. This situation is different from the study of trend changes in the returns of individual assets, where fundamenta...
Journal de Physique Lettres
Re~u le 9 aofit 1982, accepte le 4 octobre 1982)
Journal de Physique Lettres, 1982
Re~u le 12 mars 1982, accepte le 18 mai 1982)
Journal de Physique Lettres, 1982
Re~u le 12 mars 1982, accepte le 18 mai 1982)
HAL (Le Centre pour la Communication Scientifique Directe), May 31, 2021
The aim of the present study is to detect abrupt trend changes in the mean of a multidimensional ... more The aim of the present study is to detect abrupt trend changes in the mean of a multidimensional sequential signal. Directly inspired by papers of Fernhead and Liu ([4] and [5]), this work describes the signal in a hierarchical manner : the change dates of a time segmentation process trigger the renewal of a piece-wise constant emission law. Bayesian posterior information on the change dates and emission parameters is obtained. These estimations can be revised online, i.e. as new data arrive. This paper proposes explicit formulations corresponding to various emission laws, as well as a generalization to the case where only partially observed data are available. Practical applications include the returns of partially observed multi-asset investment strategies, when only scant prior knowledge of the movers of the returns is at hand, limited to some statistical assumptions. This situation is different from the study of trend changes in the returns of individual assets, where fundamental exogenous information (news, earnings announcements, controversies, etc.) can be used.
Pénalités minimales pour la sélection de modèle Mots Clefs : moindres carrés pénalisés, sélection... more Pénalités minimales pour la sélection de modèle Mots Clefs : moindres carrés pénalisés, sélection de modèle, pénalité minimale, segmentation de signal gaussien, estimation de densité, contraste pénalisé, détection de ruptures multiples, CART, arbres de régression Résumé : Dans le cadre de la sélection de modèle par contraste pénalisé, L. Birgé et P. Massart ont prouvé que le phénomène de pénalité minimale se produit pour la sélection libre parmi des variables gaussiennes indépendantes. Nous étendons certains de leurs résultats à la partition d'un signal gaussien lorsque la famille de partitions envisagée est susamment riche, notamment dans le cas des arbres de régression. Nous montrons que le même phénomène se produit dans le cadre de l'estimation de densité. La richesse de la famille de modèles est liée à une forme d'isotropie. De ce point de vue le phénomène de pénalité minimale est intrinsèque. Pour corroborer et illustrer ce point de vue, nous montrons que le même phénomène se produit pour une famille de modèles d'orientation aléatoire uniforme.
The aim of the present study is to detect abrupt trend changes in the mean of a multidimensional ... more The aim of the present study is to detect abrupt trend changes in the mean of a multidimensional sequential signal. Directly inspired by papers of Fernhead and Liu ([4] and [5]), this work describes the signal in a hierarchical manner : the change dates of a time segmentation process trigger the renewal of a piece-wise constant emission law. Bayesian posterior information on the change dates and emission parameters is obtained. These estimations can be revised online, i.e. as new data arrive. This paper proposes explicit formulations corresponding to various emission laws, as well as a generalization to the case where only partially observed data are available. Practical applications include the returns of partially observed multi-asset investment strategies, when only scant prior knowledge of the movers of the returns is at hand, limited to some statistical assumptions. This situation is different from the study of trend changes in the returns of individual assets, where fundamenta...
Dans le cadre de la selection de modele par contraste penalise, L. Birge and P. Massart ont prouv... more Dans le cadre de la selection de modele par contraste penalise, L. Birge and P. Massart ont prouve que le phenomene de penalite minimale se produit pour la selection libre parmi des variables gaussiennes independantes. Nous etendons certains de leurs resultats a la partition d'un signal gaussien lorsque la famille de partitions envisagees est suffisamment riche, notamment dans le cas des arbres de regression. Nous montrons que le meme phenomene se produit dans le cadre de l'estimation de densite. La richesse de la famille de modele s'apparente a une forme d'isotropie. De ce point de vue le phenomene de penalite minimale est intrinseque. Pour corroborer et illustrer ce point de vue, nous montrons que le meme phenomene se produit pour une famille de modeles d'orientation aleatoire uniforme.
regression tree in N (∼phylogenetic) [ (d− 1) (√ d+ √ d− 1 )2]D Lem. B.3 free line segmentation [... more regression tree in N (∼phylogenetic) [ (d− 1) (√ d+ √ d− 1 )2]D Lem. B.3 free line segmentation [ e n−1 D−1 ]D−1 1 Sec. B.2 regular binary in N 8 2 log(2) ≤ 1.39 Lem. B.4 free binary in N [( 3 + 2 √ 2 ) e n D ]D 2 Lem. B.6 regular quad in N [ 4 4 3 3 ]D 5 3 log(2)− log(3) ≤ 0.057 Lem. B.5 free quad in N [ 4e n D ]D 2 Lem. B.7 Table 16 Complexity of some types of recursive segmentation trees B.4 Enumerating the models 221
ArXiv, 2021
The aim of the present study is to detect abrupt trend changes in the mean of a multidimensional ... more The aim of the present study is to detect abrupt trend changes in the mean of a multidimensional sequential signal. Directly inspired by papers of Fernhead and Liu ([4] and [5]), this work describes the signal in a hierarchical manner : the change dates of a time segmentation process trigger the renewal of a piece-wise constant emission law. Bayesian posterior information on the change dates and emission parameters is obtained. These estimations can be revised online, i.e. as new data arrive. This paper proposes explicit formulations corresponding to various emission laws, as well as a generalization to the case where only partially observed data are available. Practical applications include the returns of partially observed multi-asset investment strategies, when only scant prior knowledge of the movers of the returns is at hand, limited to some statistical assumptions. This situation is different from the study of trend changes in the returns of individual assets, where fundamenta...
The aim of the present study is to detect abrupt trend changes in the mean of a multidimensional ... more The aim of the present study is to detect abrupt trend changes in the mean of a multidimensional sequential signal. Directly inspired by papers of Fernhead and Liu ([4] and [5]), this work describes the signal in a hierarchical manner : the change dates of a time segmentation process trigger the renewal of a piece-wise constant emission law. Bayesian posterior information on the change dates and emission parameters is obtained. These estimations can be revised online, i.e. as new data arrive. This paper proposes explicit formulations corresponding to various emission laws, as well as a generalization to the case where only partially observed data are available. Practical applications include the returns of partially observed multi-asset investment strategies, when only scant prior knowledge of the movers of the returns is at hand, limited to some statistical assumptions. This situation is different from the study of trend changes in the returns of individual assets, where fundamenta...
Journal de Physique Lettres
Re~u le 9 aofit 1982, accepte le 4 octobre 1982)
Journal de Physique Lettres, 1982
Re~u le 12 mars 1982, accepte le 18 mai 1982)
Journal de Physique Lettres, 1982
Re~u le 12 mars 1982, accepte le 18 mai 1982)