PTSD-MDNN : Fusion tardive de réseaux de neurones profonds multimodaux pour la détection du trouble de stress post-traumatique

TL;DR

PTSD-MDNN is presented, which merges two unimodal convolutional neural networks and which gives low detection error rate and could be used in the configuration of teleconsultation sessions, in the optimization of patient journeys or for human-robot interaction.

Abstract

In order to provide a more objective and quicker way to diagnose post-traumatic stress disorder (PTSD), we present PTSD-MDNN which merges two unimodal convolutional neural networks and which gives low detection error rate. By taking only videos and audios as inputs, the model could be used in the configuration of teleconsultation sessions, in the optimization of patient journeys or for human-robot interaction.

Figures (2)

• Figure 1: L'architecture de PTSD-MDNN
• Figure 2: Les convolutions spatiales et temporelles factorisées d'une convolution (2+1)D avec une taille de noyau $(3 \times 3 \times 3)$ nécessitent des matrices de poids de taille $(9 \times canaux^2) + (3 \times canaux^2)$. Ceci est moins de la moitié de celles nécessaires pour la convolution 3D complète $(27 \times canaux^2)$