Artificial Neural Networks have dramatically improved performance for many machine learning tasks. We demonstrate a new architecture for a fully optical neural network that enables a computational speed enhancement of at least two orders of magnitude and three orders of magnitude in power efficiency over state-of-the-art electronics.

Deep learning with coherent nanophotonic circuits

Data inconsistency leads to a slow training process when deep neural networks are used for the inverse design of photonic devices, an issue that arises from the fundamental property of nonuniqueness in all inverse scattering problems. Here we show that by combining forward modeling and inverse design in a tandem architecture, one can overcome this fundamental issue, allowing deep neural networks to be effectively trained by data sets that contain nonunique electromagnetic scattering instances. This paves the way for using deep neural networks to design complex photonic structures that require large training data sets.

Training Deep Neural Networks for the Inverse Design of Nanophotonic Structures

As many as 5000 holograms of high-resolution images (320 × 220 pixels) have been stored within a single crystal of Fe:LiNbO3. The nominal angle between the reference beam and the object beam was 90°. Fractal-space multiplexing was used to increase the storage bandwidth.

Angle-multiplexed storage of 5000 holograms in lithium niobate.

An approach for restoration of gray level images degraded by a known shift invariant blur function and additive noise is presented using a neural computational network. A neural network model is used to represent a possibly nonstationary image whose gray level function is the simple sum of the neuron state variables. The restoration procedure consists of two stages: estimation of the parameters of the neural network model and reconstruction of images. Owing to the model's fault-tolerant nature and computation capability, a high-quality image is obtained using this approach. A practical algorithm with reduced computational complexity is also presented. A procedure for learning the blur parameters from prototypes of original and degraded images is outlined. >

Image restoration using a neural network

Artificial intelligence tasks across numerous applications require accelerators for fast and low-power execution. Optical computing systems may be able to meet these domain-specific needs but, despite half a century of research, general-purpose optical computing systems have yet to mature into a practical technology. Artificial intelligence inference, however, especially for visual computing applications, may offer opportunities for inference based on optical and photonic systems. In this Perspective, we review recent work on optical computing for artificial intelligence applications and discuss its promise and challenges.

Inference in artificial intelligence with deep optics and photonics.

Optical implementation of content addressable associative memory based on the Hopfield model for neural networks and on the addition of nonlinear iterative feedback to a vector–matrix multiplier is described. Numerical and experimental results presented show that the approach is capable of introducing accuracy and robustness to optical processing while maintaining the traditional advantages of optics, namely, parallelism and massive interconnection capability. Moreover a potentially useful link between neural processing and optics that can be of interest in pattern recognition and machine vision is established.

/pdf/optical-implementation-of-the-hopfield-model-4f5of4v91q.pdf

Optical implementation of the Hopfield model.

Optical implementation of content addressable associative memory based on the Hopfield model for neural networks and on the addition of nonlinear iterative feedback to a vector-matrix multiplier is described. Numerical and experimental results presented show that the approach is capable of introducing accuracy and robustness to optical processing while maintaining the traditional advantages of optics, namely, parallelism and massive interconnection capability. Moreover a potentially useful link between neural processing and optics that can be of interest in pattern recognition and machine vision is established.

Aluizio Prata

Papers

Optical implementation of the Hopfield model.

Optical implementation of the Hopfield model