Stage : Modèle de Fondation Multi-Capteurs pour la Santé - (H/F)

Your responsibilities will be to explore recent scientific literature on Self-Supervised Learning, implement and train a foundation model architecture, develop robust pre-training strategies, and evaluate the relevance of learned representations.

Requirements

• State-of-the-Art: Explore recent scientific literature on Self-Supervised Learning, Transformers applied to time series, and multimodal architectures.
• Design & Development: Implement and train a foundation model architecture using Withings' historical data.
• Scientific Experimentation: Develop robust pre-training strategies to handle noisy and misaligned signals.
• Fine-tuning & Validation: Evaluate the relevance of learned representations