CoDeFel
Control for Deep and Federated Learning
Industria
Inteligencia artificial
https://cmc.deusto.eus/codefel/ ←Volver a Proyectos de investigación
Sector
Industria
Capacidad
Inteligencia artificial - Inteligencia artificial
Foco investigación UD
Transformación digital
https://cmc.deusto.eus/codefel/
Financiación
This project has received funding from the European Union’s Horizon Europe Research and Innovation Programme under Grant Agreement Nº 101096251
Objetivo
The CoDeFeL project aims to contribute to the analytical foundations of Machine Learning using Control Theory. Its objectives include developing minimal complexity ResNets, Federated Learning methodologies with privacy guarantees, and adaptive systems capable of learning in real-time, using data to optimize their dynamic behavior.
What was our contribution?
CoDeFeL focuses on several key milestones.
1. Inspired by the «Turnpike» principle, it aims to build simplified ResNet architectures and design efficient training strategies that achieve optimal performance with fewer computational resources.
2. Applying a Control Theory perspective, it seeks to develop new Federated Learning methodologies with mathematical rigor and solid statistical foundations, ensuring privacy preservation.
3. By integrating Control Theory and Machine Learning, it aims to design efficient, robust, and adaptive systems that dynamically adjust their behavior, ensuring optimal performance in dynamic and uncertain environments.
Proyectos relacionados
IndustriaACTIVO4TWINOPTIMIZACIÓN DEL PROCESO DE FABRICACIÓN DE FORJA MEDIANTE EL DESARROLLO DE UN GEMELO HÍBRIDO DIGITAL CONSIDERANDO LA SOSTENIBILIDAD Y USO…Leer másIndustriaACTIVOMEDINSPAINMétodos innovadores y cadena de valor nacional para la circularidad de sistemas de almacenamiento energético, incluyendo clasificación, reúso,…Leer másIndustriaACTIVOMOSINCOMonitorización sin contacto de composites desde su fabricación hasta su fin de vidaLeer másIndustriaACTIVOSONETOLa red social de los ActivosLeer másIndustriaACTIVOBEACONcyBErsecure industriAl Computing cONtinuumLeer másIndustriaACTIVOCoDeFelControl for Deep and Federated LearningLeer más