Photonic Extreme Learning Machine: from neuromorphic computing to universal optical interpolant, strain gauge sensor and cancer morphodynamic monitor


Funded by: MUR  
Calls: PRIN2017
Start date: 2019-08-29  End date: 2023-08-28
Total Budget: EUR 832.000,00  INO share of the total budget: EUR 261.174,00
Scientific manager: Lorenzo Pavesi   and for INO is: Gagliardi Gianluca

Organization/Institution/Company main assignee: Università di trento

other Organization/Institution/Company involved:

other INO’s people involved:

Avino Saverio
Giorgini Antonio

Abstract: The PELM project aims at developing new photonic devices for “machine learning”. Within a single neuromorphic computing architecture, different platforms are used and optimized for certain tasks based on specific characteristics. Starting from a common algorithm, through a neural network, highly innovative principle systems are created such as: sequence of optical resonators with micro-rings integrated in silicon, metasurfaces of nanowires for the synthesis of arbitrary optical functions, liquid “whispering gallery” resonators, biological resonator spheres. The unique combination of the different devices with the neural network and the light-matter interaction effects (scattering, polarization variation, absorption, resonances) allow for the creation of highly sensitive “learning machine” prototypes, opening new frontiers for AI applications in diagnostics bio-medical, autonomous driving and telecommunications.