An all-optical Diffractive Deep Neural Network (D.sup.2NN) architecture learns to implement various functions or tasks after deep learning-based design of the passive diffractive or reflective substrate layers that work collectively to perform the desired function or task. This architecture was successfully confirmed experimentally by creating 3D-printed D.sup.2NNs that learned to implement handwritten classifications and lens function at the terahertz spectrum. This all-optical deep learning framework can perform, at the speed of light, various complex functions and tasks that computer-based neural networks can implement, and will find applications in all-optical image analysis, feature detection and object classification, also enabling new camera designs and optical components that can learn to perform unique tasks using D.sup.2NNs. In alternative embodiments, the all-optical D.sup.2NN is used as a front-end in conjunction with a trained, digital neural network back-end.

Methods and apparatus for training a matrix-based differentiable program using a photonics-based processor. The matrix-based differentiable program includes at least one matrix-valued variable associated with a matrix of values in a Euclidean vector space. The method comprises configuring components of the photonics-based processor to represent the matrix of values as an angular representation, processing, using the components of the photonics-based processor, training data to compute an error vector, determining in parallel, at least some gradients of parameters of the angular representation, wherein the determining is based on the error vector and a current input training vector, and updating the matrix of values by updating the angular representation based on the determined gradients.

The invention is directed to solving an optimization problem. The method operates a photonic crossbar array structure including N input lines and M output lines, which are interconnected at junctions via N×M photonic memory devices, where N≥2 and M≥2. The photonic memory devices are programmed to store respective weights in accordance with the optimization problem. The photonic crossbar array structure is operated as follows. First, the method determines values of L input vectors of N components each, where L≥2. Second, based on the determined values, N electromagnetic signals are generated, where each of the generated signals multiplexes L input signals encoded at respective wavelengths, so as for the N electromagnetic signals to map the L input vectors of N components each. Third, the N electromagnetic signals generated are applied to the N input lines of the photonic crossbar array structure.

A photonics-assisted platform for time series prediction and classification that performs signal processing directly after the signal acquisition before any analog-to-digital conversion by using a hardware neural network with recurrent connections, implemented in a silicon photonic chip. This neural network recurrency can be implemented in silicon photonics with a much lower latency than state-of-the-art electronic systems. The recurrent neural network can detect temporal correlations and extract features from the time series signal, and therefore reduce the latency constraints for the analog-to-digital conversion and further digital signal processing.

A neuromimetic circuit includes: a primary single photon optoelectronic neuron; a synapse in optical communication with the primary single photon optoelectronic neuron; and an axonic waveguide in optical communication with the primary single photon optoelectronic neuron and the synapse such that the axonic waveguide optically interconnects the primary single photon optoelectronic neuron and the synapse.

Systems and methods for performing signed matrix operations using a linear photonic processor are provided. The linear photonic processor is formed as an array of first amplitude modulators and second amplitude modulators, the first amplitude modulators configured to encode elements of a vector into first optical signals and the second amplitude modulators configured to encode a product between the vector elements and matrix elements into second optical signals. An apparatus may be used to implement a signed value of an output of the linear processor. The linear photonic processor may be configured to perform matrix-vector and/or matrix-matrix operations.

Our high phase sensitivity wide-field phase cancellation interferometry system allows single-shot, label-free optical sensing of neural action potentials via imaging of optical path length changes. Single-shot sensing and monitoring of single neurons within a neural network should lead to a more comprehensive understanding neural network processing, which is beneficial for the advancement in the field of neuroscience as well as its biomedical applications and impact. Our system cancels the phase profile of the resting neuron from the phase profile of the spiking neuron, improving the sensitivity by two orders of magnitude. Using a detector with an extremely large well depth and an appropriately biased interferometer increases the sensitivity by another order of magnitude, yielding a measurement that is three orders of magnitude more sensitive than those possible with other microscopes.

Disclosed are coherent photonic circuit architectures that optically implement linear algebraic computations. In neuromorphic applications of such photonic circuit architectures, individual neural network layers can be implemented by coherent optical linear neurons in a crossbar configuration, integrated with electronic circuitry at the interfaces between neural network layers to determine the neuron inputs to one layer based on the neuron outputs of the preceding layer. Wavelength division multiplexing can be used to efficiently implement certain specific network models, optionally in conjunction with electro-optic switches to render a generic hardware configuration programmable.

Embodiments are disclosed for computing an eye diagram based on input pulse responses. An example method includes receiving a set of input pulse responses in one or more unit interval (UI) spaced samples. The set of input pulse responses is generated based on measuring a signal histogram of a receiver of a pulse amplitude modulation analog signal. The method further includes receiving a set of voltage range constraints and generating a matrix based at least in part on an element-wise trigonometric-based operation performed on one or more products of each element of the set of input pulse responses and the set of voltage range constraints. The method further includes generating an eye diagram probability density function based on the matrix and computing an eye diagram based on the eye diagram probability density function, the voltage range constraints, and time data associated with the one or more unit interval spaced samples.

An optoelectronic synaptic memristor includes: a bottom electrode layer, a porous structure layer modified with quantum dots, a two-dimensional material layer, a transparent top electrode layer, and a waveguide layer, which are arranged in sequence from top to bottom, wherein the waveguide is ridge shaped for light conduction, comprising a wedge-shaped output terminal, wherein: through the wedge-shaped output terminal of the waveguide, light is vertically injected into the two-dimensional material layer and the porous structure layer modified with the quantum dots. By integrating the waveguide and the optoelectronic memristor, the present invention obtains the highly controlled characteristics with high alignment and confinement for light effect on the device and has advantages in realizing optoelectronic synergy in the optoelectronic synaptic memristors. The present invention has strong controllability and excellent performance and can be widely used in high-density integration of storage and computing, artificial synapses, artificial intelligence, etc.