Vers une nouvelle Interaction Homme Environnement dans les jeux vidéo et pervasifs : rétroaction biologique et états émotionnels. Apprentissage profond non supervisé au service de l’affectique.

Viviane Gal 1
1 CEDRIC - ILJ - CEDRIC - Interactivité pour Lire et Jouer
CEDRIC - Centre d'études et de recherche en informatique et communications
Abstract : Again and again adapting to software and games is the fate of users and players. Our approach is different and wants to place the human in the heart of space, in its context. Adapting video games or pervasives to the player according to his emotions is a challenge. Proposing a new Human Environment Interaction in games with feedback related to emotional states raises many questions especially when the states come from physiological information measured remotely. It is a real challenge for the affectics. Indeed, we have to start by wondering whether it is possible to discover emotional states from contact biosensors measurements. This is one of the first questions we answer using unsupervised data analysis techniques, automatic and deep learning, based on Kohonen maps and a hybrid EMDeep model, a combination of a Gaussian mixture model, a maximization expectation algorithm and a regression method. All the analyzes are done blindly. We do not point out what the data can reveal. These are the data that express themselves. In both cases we come to a classification of individuals’ types, or even a recognition of states. This recognition prepares our standardization, the reference we are trying to build. The states are not identified yet. To do so, they require a deepening that we have already engaged using other methods such as fuzzy logic. Another question raised, is it possible to replace contact biosensors with remote, non-invasive devices, obtaining the same results? The answer is incomplete because it is necessary to refine our algorithms and to renew the experiments to improve our model even to finalize it. Nevertheless, we propose processes that fulfill the function of remote capture that will have to be tested with the calibration that we propose. The techniques developed propose classification solutions adapted to the data of the electric signal type. The reflection on an anthropocentric interaction brings a singular point of view on a new way of conceiving a network of human communication environment and human human.
Complete list of metadatas

Cited literature [190 references]  Display  Hide  Download

https://hal-cnam.archives-ouvertes.fr/tel-02477340
Contributor : Viviane Gal <>
Submitted on : Thursday, February 13, 2020 - 12:12:06 PM
Last modification on : Saturday, February 15, 2020 - 1:28:07 AM

File

These_Viviane_Gal .pdf
Files produced by the author(s)

Identifiers

  • HAL Id : tel-02477340, version 1

Collections

Citation

Viviane Gal. Vers une nouvelle Interaction Homme Environnement dans les jeux vidéo et pervasifs : rétroaction biologique et états émotionnels. Apprentissage profond non supervisé au service de l’affectique.. Informatique. Conservatoire National des Arts et Métiers, 2019. Français. ⟨tel-02477340⟩

Share

Metrics

Record views

34

Files downloads

6