Modélisation des ondes de pression transitoires dans les réseaux de distribution d’eau

Autor: Bayle, Alexandre

Dans les zones urbaines, la disponibilité de l’eau potable repose souvent sur des systèmes de distribution complexes, dont la dynamique est encore mal comprise. La supervision des réseaux, qui s’inscrit dans le prolongement des projets de "sma

Externí odkaz: https://explore.openaire.eu/search/publication?articleId=dedup_wf_001::c9ec10e6ea5ac952538b57c0c2c0c686
https://oatao.univ-toulouse.fr/29640/

Apprentissage de circuits quantiques par descente de gradient classique

Autor: Lamarre, Aldo

Nous présentons un nouvel algorithme d’apprentissage de circuits quantiques basé sur la descente de gradient classique. Comme ce sujet unifie deux disciplines, nous expliquons les deux domaines aux gens de l’autre discipline. Conséquemment, no

Externí odkaz: http://hdl.handle.net/1866/24322

Optimal Memory-aware Backpropagation of Deep Join Networks

Autor: Beaumont, Olivier, Herrmann, Julien, Pallez, Guillaume, Shilova, Alena

Publikováno v: [Research Report] RR-9273, Inria. 2019

In the context of Deep Learning training, memory needs to store activations can prevent the user to consider large models and large batch sizes. A possible solution is to rely on model parallelism to distribute the weights of the model and the activa

Externí odkaz: https://explore.openaire.eu/search/publication?articleId=od_______212::52e23121713afd856170859a56ded03e
https://hal.inria.fr/hal-02131552/document

Rétropropagation optimale de réseau profonds à forme de Joins

Autor: Guillaume Pallez, Olivier Beaumont, Alena Shilova, Julien Herrmann

Publikováno v: Philosophical Transactions of the Royal Society A: Mathematical, Physical and Engineering Sciences
Philosophical Transactions of the Royal Society A: Mathematical, Physical and Engineering Sciences, In press
[Research Report] RR-9273, Inria. 2019
Philosophical Transactions of the Royal Society A: Mathematical, Physical and Engineering Sciences, Royal Society, The, In press
Philos Trans A Math Phys Eng Sci

In the context of Deep Learning training, memory needs to store activations can prevent the user to consider large models and large batch sizes. A possible solution is to rely on model parallelism to distribute the weights of the model and the activa

Externí odkaz: https://explore.openaire.eu/search/publication?articleId=doi_dedup___::81f8fa3ba353f137063506efd4a1c7a7
https://hal.inria.fr/hal-02401105

Predictive and adaptive models for individual residential building energy management : Artificial neural networks based on commonly available data

Autor: Delorme-Costil, Alexandra

Publikováno v: Génie des procédés. Ecole des Mines d'Albi-Carmaux, 2018. Français. ⟨NNT : 2018EMAC0020⟩

The use of predictive control permits to reduce the energy consumption of residential buildings without reducing the comfort of the inhabitant. The company BoostHeat develops a thermodynamic furnace with high energy efficiency for this purpose. Simul

Externí odkaz: https://explore.openaire.eu/search/publication?articleId=od_______212::7077353702c47c4ebc4bc7e4d02a8ee7
https://tel.archives-ouvertes.fr/tel-02475997/document