Summary:
We will take a look at the AIMAC (Artificial Intelligence for Multimodal Affective Computing) research team and discuss the analysis of emotions through facial expressions, prosody, and vocalized text. We will then address the synthesis of emotions across these same three modalities (expressive 3D character animation, GPT4-style text generation) and give a brief overview of several applications, particularly in the healthcare field. We will present the research valorization policy at CentraleSupélec through the creation of four startups born from the team, with a focus on the company Emobot, which markets a tool for detecting and monitoring the evolution of signs related to mood disorders (depression, bipolar disorder). To conclude, and time permitting, we will discuss ethics and Artificial Intelligence in the field of mental health.
Biography:
Renaud Séguier is a Professor at CentraleSupélec, holding a PhD and HDR from the University of Rennes I. He leads the AIMAC (Artificial Intelligence for Multimodal Affective Computing) team within the Image department of IETR (UMR 6164). AIMAC works on latent spaces (PCA, autoencoders, GANs) and attempts to uncover manifolds within these spaces in an unsupervised manner. The applied framework concerns the multimodal analysis and synthesis of emotions (image, prosody, text) in a medical context. He has co-founded four startups directly stemming from his research: Dynamixyz (2010), 3D Sound Labs (2014), Immersive Therapy (2017), and Emobot (2022).
