Interprétation collaborative de séries temporelles. Application à des données de réanimation médicale.

Abstract : This thesis proposes an approach of the man/machine collaboration inspired by the theory of Enaction, in interpretation of complex data. The collaboration aims to enable them to make an interpretation based on their own skills. To preserve their autonomy, the communication between partners is achieved with annotations. This approach is applied to the design of a multi-agent system for the collaborative interpretation of physiological signals recorded on patients in intensive care medical, i.e. multivariate time series. According to the autonomy principle, the system is able to construct an interpretation of the time series by the construction (1) of models of events and scenarios (sets of events linked by temporal relations) and (2) of the annotations from these models. According to the principle of adaptation, the treatments and models of the system evolve to take into account the human's annotations in the course of interpretation.
Document type :
Theses
Complete list of metadatas

https://tel.archives-ouvertes.fr/tel-00264145
Contributor : Thomas Guyet <>
Submitted on : Friday, March 14, 2008 - 1:03:54 PM
Last modification on : Tuesday, September 24, 2019 - 4:22:02 PM
Long-term archiving on : Thursday, May 20, 2010 - 10:04:41 PM

Identifiers

  • HAL Id : tel-00264145, version 1

Citation

Thomas Guyet. Interprétation collaborative de séries temporelles. Application à des données de réanimation médicale.. Informatique [cs]. Institut National Polytechnique de Grenoble - INPG, 2007. Français. ⟨tel-00264145⟩

Share

Metrics

Record views

1256

Files downloads

1659