Metamaterials and machine learning photonics

Organisers

Francesco Rossella

Università di Modena e Reggio, Italy

Claudio Conti

Sapioenza Università di Roma, Italy

Francesco Floris

Tyndall National Institute, Ireland

Keywords: metamaterials nanostructures, machine learning

3 SESSIONS / 2 DAYS, with one introductory talk for each day, one/two keynote invited talks for each session, and regular talks.

OPTICAL MATERIALS such as arrays of scattering building blocks with a SUBWAVELENGTH size and periodicity provide a well-established platform for light manipulation together with extraordinary control over light–matter interaction at the nanoscale. For instance, artificial calibrated arrangements of well-aligned dielectric nanostructures acting as nanostructured optical reflectors have been proven to enable the control of polarization and the orbital angular momentum of light, dispersive holograms and spatial light modulators. The quest for novel methods to realize functional assemblies of NANOSTRUCTURES characterized by conveniently tailored optical responses ceaselessly triggers the interest of scientists from many different areas: in this context, semiconductor nanowires (NWs) emerge as a very promising class of quasi one-dimensional systems ideally suited for building up photonic platforms, with huge potential of NW-based systems for optical response engineering and light manipulation.

Figure 1: Vogel spiral made of semiconductor nanowires (courtesy of F.Rossella and L- Sorba)

On the other hand, PHOTONICS emerges as the new pathway for low power, efficient and fast MACHINE LEARNING IMPLEMENTATION. In this context, huge investigation efforts are carried out to conceive, design and demonstrate prototypes of photonic processors implementing an EXTREME LEARNING MACHINE, i.e., a single hidden layer feed-forward neural network with randomly generated hidden nodes and analytically computed output weights. An interdisciplinary community with complementary competences in photonic integrated devices, complex systems, linear and nonlinear optical measurements, bio-physics and semiconductor technology is developing various platforms for photonic learning machines and their validation on killer applications. This mini-colloquium aims at presenting the most recent results achieved by this interdisciplinary community working on PHOTONIC DEVICES FOR MACHINE LEARNING.