Experimentally realized in situ backpropagation for deep learning in photonic neural networks.

Integrated photonic neural networks provide a promising platform for energy-efficient, high-throughput machine learning with extensive scientific and commercial applications. Photonic neural networks efficiently transform optically encoded inputs using Mach-Zehnder interferometer mesh networks interleaved with nonlinearities. We experimentally trained a three-layer, four-port silicon photonic neural network with programmable phase shifters and optical power monitoring to solve classification tasks using "in situ backpropagation," a photonic analog of the most popular method to train conventional neural networks. We measured backpropagated gradients for phase-shifter voltages by interfering forward- and backward-propagating light and simulated in situ backpropagation for 64-port photonic neural networks trained on MNIST image recognition given errors. All experiments performed comparably to digital simulations ([Formula: see text]94% test accuracy), and energy scaling analysis indicated a route to scalable machine learning.

Experimental realization of arbitrary activation functions for optical neural networks.

We experimentally demonstrate an on-chip electro-optic circuit for realizing arbitrary nonlinear activation functions for optical neural networks (ONNs). The circuit operates by converting a small portion of the input optical signal into an electrical signal and modulating the intensity of the remaining optical signal. Electrical signal processing allows the activation function circuit to realize any optical-to-optical nonlinearity that does not require amplification. Such line shapes are not constrained to those of conventional optical nonlinearities. Through numerical simulations, we demonstrate that the activation function improves the performance of an ONN on the MNIST image classification task. Moreover, the activation circuit allows for the realization of nonlinearities with far lower optical signal attenuation, paving the way for much deeper ONNs.

Wave physics as an analog recurrent neural network.

Analog machine learning hardware platforms promise to be faster and more energy efficient than their digital counterparts. Wave physics, as found in acoustics and optics, is a natural candidate for building analog processors for time-varying signals. Here, we identify a mapping between the dynamics of wave physics and the computation in recurrent neural networks. This mapping indicates that physical wave systems can be trained to learn complex features in temporal data, using standard training techniques for neural networks. As a demonstration, we show that an inverse-designed inhomogeneous medium can perform vowel classification on raw audio signals as their waveforms scatter and propagate through it, achieving performance comparable to a standard digital implementation of a recurrent neural network. These findings pave the way for a new class of analog machine learning platforms, capable of fast and efficient processing of information in its native domain.

Design of a tapered slot waveguide dielectric laser accelerator for sub-relativistic electrons.

We propose a dielectric laser accelerator design based on a tapered slot waveguide structure for sub-relativistic electron acceleration. This tapering scheme allows for straightforward tuning of the phase velocity of the accelerating field along the propagation direction, which is necessary for maintaining synchronization with electrons as their velocities increase. Furthermore, the non-resonant nature of this design allows for better tolerance to experimental errors. We also introduce a method to design this continuously tapered structure based on the eikonal approximation, and give a working example based on realistic experimental parameters.

Plasmonic Circuit Theory for Multiresonant Light Funneling to a Single Spatial Hot Spot.

We present a theoretical framework, based on plasmonic circuit models, for generating a multiresonant field intensity enhancement spectrum at a single "hot spot" in a plasmonic device. We introduce a circuit model, consisting of an array of coupled LC resonators, that directs current asymmetrically in the array, and we show that this circuit can funnel energy efficiently from each resonance to a single element. We implement the circuit model in a plasmonic nanostructure consisting of a series of metal bars of differing length, with nearest neighbor metal bars strongly coupled electromagnetically through air gaps. The resulting nanostructure resonantly traps different wavelengths of incident light in separate gap regions, yet it funnels the energy of different resonances to a common location, which is consistent with our circuit model. Our work is important for a number of applications of plasmonic nanoantennas in spectroscopy, such as in single-molecule fluorescence spectroscopy or Raman spectroscopy.