Adel Taleb, PhD

SogetiLabs (Capgemini) - Data Scientist and Head of Data Department

September 2020 - October 2024

At SogetiLabs, I led several large-scale projects as the head of the data department. My key responsibilities included:

• Motor behavior classification project for newborns: Developed an AI-based model to detect abnormalities in general movements (GMs) of infants, enabling earlier detection of neuro-motor disorders.
• Deep learning model optimization: Implemented a pipeline to optimize the performance of deep learning models on medical datasets by adapting convolutional neural network architectures.
• Team supervision: Supervised and mentored a team of data scientists, guiding interns and consultants in AI-focused R&D projects.

CHART Laboratory (PSL) - PhD Student

September 2019 - October 2024

During my PhD at PSL, I worked on several major research topics related to computer vision applied to human motor skills:

• Spinal Muscular Atrophy (SMA) classification model: Developed a classification model to diagnose SMA from infant videos, designed to be interpretable by clinicians.
• Study of pathological movements in children: Collaborated with pediatric neurologists to analyze abnormal movements in infants and adapted deep learning techniques to handle limited medical data.

SogetiLabs (Capgemini) - Data Science Intern

September 2018 - September 2019

As a data science intern, I contributed to innovative projects within the R&D team:

• Named entity recognition (NER): Developed deep learning algorithms for entity recognition in text documents and their anonymization.
• Predictive data analysis: Participated in predictive analysis projects using supervised and unsupervised machine learning techniques.

PhD in Computer Science, Statistics, and Cognition - Université PSL, CHArt Laboratory

2020 - 2024

Thesis title: Motor Skills Analysis in Infants and Adults

• Study of General Movements in Infants through Machine Learning: Conducted an advanced state-of-the-art study on general movements and extracted spatio-temporal descriptors using computer vision techniques.
• Adaptation of Pose Estimation Models for Infants: Optimized deep learning models for pose detection and tracking in video data with a limited number of samples.
• Development of a Classification Model for Spinal Muscular Atrophy (SMA) and XAI: Created a classification pipeline based on features extracted from movement sequences, incorporating supervised learning methods to provide pre-diagnoses from videos while considering clinical acceptability through model explainability techniques.
• Biomechanical Analysis of Gait: Introduced a new method based on dynamic movement analysis using the barycenter, applied to the automatic detection of hemiplegic gait in adults (patented method).
• Set up several data collection protocols in compliance with data protection requirements.
• Participated in the creation of the research platform R2P2.

Master's Degree in Machine Learning for Data Science - Université Paris Descartes

2017 - 2019

• Assessment of document embeddings quality (Text Mining).
• Implementation of a Generative Adversarial Network (GAN) for pseudo-labeling (Deep Learning).
• Deep learning for dimensionality reduction.
• AI4Eye Project: Designed a learning model for video image recognition and interpretation aimed at visually impaired individuals, with distance estimation (near and far).
• Developed a recommendation system based on user-based collaborative filtering, with an estimation of recommendation quality by applying different clustering methods and comparing clusters before and after recommendation.