We will discuss diffractive optical networks designed by deep learning to all-optically implement various complex functions as the input light diffracts through spatially-engineered surfaces. These diffractive processors complete their computational task at the speed of light propagation through thin, passive optical layers and have various applications, e.g., all-optical image analysis, feature detection, object classification, computational imaging and seeing through diffusers. They also enable task-specific camera designs and new optical components for, e.g., spatial, spectral and temporal beam shaping, polarization engineering and spatially-controlled wavelength division multiplexing. These deep learning-designed diffractive networks broadly impact (1) all-optical statistical inference engines, (2) computational cameras and microscopes, and (3) inverse design of optical systems that are task-specific.
|