CIM Seminar with Francesco Ferranti

Abstract: Photonic nanostructures have a tremendous potential in many application fields such as imaging, sensing, and photonic integrated circuits. The design of photonic nanostructures requires the use of electromagnetic (EM) solvers. These solvers can result very computationally expensive. A design flow needs to carry out design tasks such as optimization and variability analysis, and therefore a high number of EM simulations can be required. The direct use of EM solvers into a design flow then can easily become unfeasible.
Machine learning (ML) techniques can be used to tremendously speed-up the design flow while retaining high accuracy in the obtained results. Many very interesting and challenging directions are to be explored, where the cross-fertilization of ML techniques and EM design problems becomes the key enabler for disruptive application results.

Registration