At Adagos, we are driven by a shared belief in the power of innovation and collaboration to shape the future of artificial intelligence.
Our commitment to pushing the boundaries of technology enables us to deliver pioneering, tailor-made solutions specifically designed to meet the unique needs of engineers. Whether through our unique story, the expertise of our team, or our collective knowledge, we strive to create a lasting impact in the digital world.
During his academic career, Mohamed Masmoudi has been dedicated to developing mathematical tools designed to meet industrial needs. Among these tools, one notable contribution is the topological gradient theory, which enables the determination of optimal mechanical, fluidic, electromagnetic, or electronic structures.
Then he had the idea of applying this approach to artificial intelligence, leading to the design of neural networks that are minimal in size. This innovation ultimately resulted in the creation of Adagos and its flagship product, NeurEco.
Once again, this innovation is particularly well-suited to addressing industrial needs for the creation of digital twins of operational systems. This capability is essential for enhancing system availability and boosting performance.
Anomaly detection and predictive maintenance are driven by the drift observed between the real system’s response and that of its digital twin.
NeurEco is oriented to physical phenomena simulation and is grounded in fundamental scientific principles, with parsimony playing a key role. As Einstein stated: « Everything must be made as simple as possible, but not simpler. »
NeurEco is not merely learning the data, but learning from it, as it identifies and infers the underlying rules that structure the data.
Parsimony drives this process, leading to models so small and efficient that there’s no need to compromise between complexity and accuracy.
We do not claim that this concept works with other applications like NLP. Unlike physical phenomena, which are governed by a limited number of rules, NLP models need to be large in order to memorize thousands of exceptions to grammatical rules.
In addition to the topological gradient, Adagos’ technology is the result of two competencies:
Our tools, NeurEco Dynamic and NeurEco PFS (Parametric Frequency Sweep), stem from this expertise.
Once again, parsimony played a key role. The frequency and time domains are two sides of the same coin, both characterized by significant instability. The parsimony of NeurEco is the keystone of these products, and any deviation from this principle leads to a degradation in prediction accuracy.
Our team brings together specialists in applied mathematics, physics, artificial intelligence, and industrial engineering.
This multidisciplinary culture allows us to combine scientific rigor with practical experience, delivering efficient AI solutions that remain grounded in physical understanding.
