Design optimization of high-speed links via machine learning

Machine learning has become a mature and powerful tool for solving problems in many different areas, including computational finance, image processing, energy forecasting and voice recognition. What is more important, the users can now rely on readily available state-of-the-art algorithms which are embedded in either open source or commercial software (e.g., Matlab). The typical applications of the above techniques are however mainly related to classification problems, possibly involving heterogeneous input data. The aim of this activity is to apply this class of algorithms to the parametric analysis and design optimization of realistic electronic links (e.g., a printed circuit board structure in a smartphone). The research work attempts bridging the gap between this promising solution and standard approaches.


ERC Sector:

  • PE7_3 Simulation engineering and modelling
  • PE7_4 (Micro and nano) systems


  • Machine learning
  • Interconnected systems
  • Uncertainty
  • Design optimization

Research groups