Yujin Wu

Yujin Wu

Auto-generated profile, based on publicly available information

Yujin Wu

Univ. Lille, CNRS, Centrale Lille, UMR 9189 CRIStAL, Lille, France

Research Focus & Metrics

Academic Career Length

5Years (2022-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
3 April 2023 in IEEE Transactions on Affective Computing
Verified
Journal Article
Closed access
YWYujin Wu
Yujin Wu
Mohamed DaoudiMohamed Daoudi
Mohamed Daoudi
AAAli Amad
Ali Amad
Recently, wearable emotion recognition based on peripheral physiological signals has drawn massive attention due to its less invasive nature and its applicability in real-life scenarios. However, how to effectively fuse multimodal data remains a challenging problem. Moreover, traditional fully-supervised based approaches suffer from overfitting given limited labeled data. To address the above issues, we propose a novel self-supervised learning (SSL) framework for wearable emotion recognition, where efficient multimodal fusion is realized with temporal convolution-based modality-specific encoders and a transformer-based shared encoder, capturing both intra-modal and inter-modal correlations. Extensive unlabeled data is automatically assigned labels by five signal transforms, and the proposed SSL model is pre-trained with signal transformation recognition as a pretext task, allowing the extraction of generalized multimodal representations for emotion-related downstream tasks. For evaluation, the proposed SSL model was first pre-trained on a large-scale self-collected physiological dataset and the resulting encoder was subsequently frozen or fine-tuned on three public supervised emotion recognition datasets. Ultimately, our SSL-based method achieved state-of-the-art results in various emotion classification tasks. Meanwhile, the proposed model was proved to be more accurate and robust compared to fully-supervised methods on low data regimes.

4 February 2022 in IEEE Access
Verified
Journal Article
Open access
EGEman Ahmed Abdel Ghaffar
Eman Ahmed Abdel Ghaffar
YWYujin Wu
Yujin Wu
Mohamed DaoudiMohamed Daoudi
Mohamed Daoudi
Emotion recognition plays an important role in human computer interaction systems as it helps the computer in understanding human behavior and their decision making process. Using Electroencephalographic (EEG) signals in emotion recognition offers a direct assessment on the inner state of human mind. This study aims to build a subject dependent emotion recognition system that differentiate between high and low levels of valance and arousal, using multidimensional EEG signals. Our system offers a transfer learning- minimum distance to Riemannian mean (TL-MDRM) framework. In this work, we perform two pre-processing stages. In the first stage, we analyze the EEG signals to investigate their non-Gaussianity and determine the most appropriate signal distribution. Using several statistical and goodness of fit tests, T-distribution was found to be the most appropriate distribution. Covariance matrix estimations plays a crucial step in manifold learning technique, based on the most suitable signal distribution the covariance matrix estimation technique is chosen. In the second stage, we perform transfer learning to deal with cross-session variability by generating a unique reference point for each participant and performing affine transformation for the covariance matrices on the symmetric positive definite (SPD) manifold around this point. The results show that, TL process improved the performance even when assuming Gaussian distribution, while assuming T-distribution with TL improved the performance further.

29 March 2023 in arXiv
Verified
Preprint
Open access
PDF file
YWYujin Wu
Yujin Wu
Mohamed DaoudiMohamed Daoudi
Mohamed Daoudi
AAAli Amad
Ali Amad
Recently, wearable emotion recognition based on peripheral physiological signals has drawn massive attention due to its less invasive nature and its applicability in real-life scenarios. However, how to effectively fuse multimodal data remains a challenging problem. Moreover, traditional fully-supervised based approaches suffer from overfitting given limited labeled data. To address the above issues, we propose a novel self-supervised learning (SSL) framework for wearable emotion recognition, where efficient multimodal fusion is realized with temporal convolution-based modality-specific encoders and a transformer-based shared encoder, capturing both intra-modal and inter-modal correlations. Extensive unlabeled data is automatically assigned labels by five signal transforms, and the proposed SSL model is pre-trained with signal transformation recognition as a pretext task, allowing the extraction of generalized multimodal representations for emotion-related downstream tasks. For evaluation, the proposed SSL model was first pre-trained on a large-scale self-collected physiological dataset and the resulting encoder was subsequently frozen or fine-tuned on three public supervised emotion recognition datasets. Ultimately, our SSL-based method achieved state-of-the-art results in various emotion classification tasks. Meanwhile, the proposed model proved to be more accurate and robust compared to fully-supervised methods on low data regimes.

7 April 2026
Verified
Thesis
Closed access
YWYujin Wu
Yujin Wu
Reconnaissance multimodale des émotions à partir de signaux physiologiques et d'expressions faciales: La reconnaissance des émotions est un sous-domaine de l'informatique affective et une direction de recherche critique pour le développement de l'intelligence artificielle centrée sur l'homme, qui vise l'identification et l'interprétation des émotions humaines par des machines de manière objective et automatique.Les émotions peuvent être exprimées et transmises par divers canaux qui peuvent être classés en deux grandes catégories : les modalités comportementales observables de l'extérieur, telles que les expressions faciales, les postures corporelles et la parole, et les modalités physiologiques intrinsèques, telles que les activités cérébrales, la conductivité de la peau, le rythme cardiaque, etc. Des méthodes de reconnaissance des émotions ont donc été développées sur la base de ces modalités.Les récentes avancées en matière de technologie des capteurs et de systèmes d'interaction homme-machine offrent la possibilité de faire passer le déploiement des systèmes de reconnaissance des émotions du laboratoire au monde réel. Cependant, la plupart des recherches existantes sont encore orientées vers le premier.Dans ce contexte, nous nous intéressons à la conception d'algorithmes de reconnaissance des émotions pour des scénarios de la vie réelle d'un point de vue physiologique, comportemental et multimodal et à l'exploration des implications et de l'impact de leurs interventions sur des applications pratiques.Dans cette thèse, du point de vue physiologique, nous proposons une méthode d'apprentissage de représentation multimodale auto-supervisée pour la reconnaissance d'émotion portable basée sur des signaux physiologiques périphériques pour faire face au problème du surapprentissage posé par des données étiquetées limitées et le biais d'étiquetage inexact dans le monde réel. Du point de vue comportemental, nous suggérons une méthode d'apprentissage métrique non euclidienne pour la reconnaissance d'expressions basées sur des points de repère faciaux en 2D afin de résoudre des problèmes tels que la variation de pose et la variation intra-classe.D'un point de vue multimodal, nous présentons un cadre géométrique profond basé sur une représentation matricielle symétrique définie positive pour la détection multimodale du stress et de la douleur afin de résoudre le problème de la fusion des modalités physiologiques et comportementales. En outre, nous réalisons le déploiement de la reconnaissance des émotions dans un scénario réel, à savoir la formation par simulation médicale, où l'algorithme proposé précédemment est intégré dans une interface utilisateur graphique pour tester sa faisabilité et sa validité sur les données multimodales collectées, tout en explorant ses effets pédagogiques.

14 December 2022
Verified
Book Chapter
Closed access
PDF file
YWYujin Wu
Yujin Wu
Mohamed DaoudiMohamed Daoudi
Mohamed Daoudi