News

A model that integrates the underlying physics will generalize well, even with limited data. A purely correlative model, on the other hand, needs large datasets to compensate for what it doesn’t understand. That’s why we design neural networks that are parsimonious by construction: they adapt to the structure of the problem rather than just fitting the data. 🔎 The result: lighter, more interpretable, and more robust predictive models, the very philosophy behind NeurEco, our

September will be a month of new opportunities to discover, learn, and connect:  🔹 An important evolution in our AI technology designed...

GDTech and Adagos enter a strategic partnership to accelerate the use of AI in engineering, starting with the integration of NeurEco into...

Le 23 septembre 2025, nous vous invitons à une matinée exclusive à Wellio Lyon (9 rue des Cuirassiers, 69003 Lyon), dès 9h00. 📢 Au...

Ornithopter - Airplane - Neural Network Just like early aviation attempted to mimic birds flapping their wings before discovering...

Parsimonious neural network Many AI solutions turn to TinyML or AutoML to shrink large models or automate training. But these approaches...

In many industrial applications, it's possible to build reliable digital twins with 10 to 100 times less data than commonly expected....