A system may include an optical source and an adjustable spatial light modulator coupled to the optical source. The system may further include a medium coupled to the adjustable spatial light modulator, and an optical detector coupled to the medium. The optical detector may obtain various optical signals that are transmitted through the medium at various predetermined spatial light modulations using the adjustable spatial light modulator. The system may further include a controller coupled to the optical detector and the adjustable spatial light modulator. The controller may train an electronic model using various synthetic gradients based on the optical signals.

A sensor system includes an array of optical sensors on an integrated circuit and a plurality of sets of optical filters atop at least a portion of the array. Each set of optical filters is associated with a set of optical sensors of the array, with a set of optical filters including a plurality of optical filters, with each optical filter being configured to pass light in a different wavelength range. A first interface is configured to interface with the optical sensors and first processing circuitry that is configured to execute operational instructions for receiving an output signal representative of received light from the optical sensors and determining a spectral response for each set of optical sensors. A second interface is configured to interface with the first processing circuitry with second processing circuitry that is configured for determining, based on the spectral response for each set of optical sensors, an illuminant spectrum for each spectral response and then substantially remove the illuminant spectrum from the spectral response.

A sensor system includes an array of optical sensors on an integrated circuit and a plurality of sets of optical filters atop at least a portion of the array. Each set of optical filters is associated with a set of optical sensors of the array, with a set of optical filters including a plurality of optical filters, with each optical filter being configured to pass light in a different wavelength range. A first interface is configured to interface with the optical sensors and first processing circuitry that is configured to execute operational instructions for receiving an output signal representative of received light from the optical sensors and determining a spectral response for each set of optical sensors. A second interface is configured to interface with the first processing circuitry with second processing circuitry that is configured for determining, based on the spectral response for each set of optical sensors, an illuminant spectrum for each spectral response and then substantially remove the illuminant spectrum from the spectral response.

An electronic device includes a memory that stores input matrices A and B, a cache memory, and a processor. The processor generates a compiled representation that includes values for acquiring data from input matrix A when processing instances of input data through the neural network, the values including a base address in input matrix A for each thread from among a number of threads and relative offsets, the relative offsets being distances between elements of input matrix A to be processed by the threads. The processor then stores, in the local cache memory, the compiled representation including the base address for each thread and the relative offsets.

An optical computing device including a parametric oscillator array, an interaction computing matrix, a first feedback module connected to two ends of the parametric oscillator array, and a second feedback module connected to the parametric oscillator array and the interaction computing array. The parametric oscillator array is configured to receive a first group of signals, and generate a first group of optical signals including a plurality of first optical signals. The interaction computing array is configured to receive the first group of optical signals, and perform matrix operation on the first group of optical signals. The first feedback module is configured to receive the first group of optical signals, and transmit the first group of optical signals to the parametric oscillator array. The second feedback module is configured to receive the second group of optical signals, and transmit the second group of optical signals to the parametric oscillator array.

A manufacturing method for a photonic device includes dividing a target photonic network, which is a photonic network configured for the photonic semiconductor device, into a plurality of sub-photonic networks, forming the plurality of sub-photonic networks on a plurality of photonic chips, and connecting the plurality of sub-photonic networks on the plurality of photonic chips through a coupler to obtain the photonic semiconductor device carrying the target photonic network, wherein the coupler is configured to couple light from one photonic chip to another photonic chip. Compared with the scale of the photonic network of the existing photonic semiconductor device, which is limited due to the footprint limitation of a single chip, the scale of the photonic network of the photonic semiconductor device is increased several times.

An interface structure for a biological environment including at least one composite electrical impulse generating layer comprising a matrix phase of a piezo polymer material, a first dispersed phase of piezo nanocrystals, and second dispersed phase of carbon nanotubes, the first and second dispersed phase presented through the matrix phase. The piezo polymer material and piezo nanocrystal convert mechanical motion into electrical impulses and accept electrons to charge the composite impulse generating layer. The carbon nanotubes provide pathways for distribution of the electrical impulses to a surface of the composite impulse generating layer contacting the biological environment. The carbon nanotubes further provide for the delivery of the byproducts of the free radical degradation from the biological environment to both piezo-nanocrystals and piezo-polymer.

Deep learning performance is limited by computing power, which is in turn limited by energy consumption. Optics can make neural networks faster and more efficient, but current schemes suffer from limited connectivity and the large footprint of low-loss nanophotonic devices. Our optical neural network architecture addresses these problems using homodyne detection and optical data fan-out. It is scalable to large networks without sacrificing speed or consuming too much energy. It can perform inference and training and work with both fully connected and convolutional neural-network layers. In our architecture, each neural network layer operates on inputs and weights encoded onto optical pulse amplitudes. A homodyne detector computes the vector product of the inputs and weights. The nonlinear activation function is performed electronically on the output of this linear homodyne detection step. Optical modulators combine the outputs from the nonlinear activation function and encode them onto optical pulses input into the next layer.

Deep learning performance is limited by computing power, which is in turn limited by energy consumption. Optics can make neural networks faster and more efficient, but current schemes suffer from limited connectivity and the large footprint of low-loss nanophotonic devices. Our optical neural network architecture addresses these problems using homodyne detection and optical data fan-out. It is scalable to large networks without sacrificing speed or consuming too much energy. It can perform inference and training and work with both fully connected and convolutional neural-network layers. In our architecture, each neural network layer operates on inputs and weights encoded onto optical pulse amplitudes. A homodyne detector computes the vector product of the inputs and weights. The nonlinear activation function is performed electronically on the output of this linear homodyne detection step. Optical modulators combine the outputs from the nonlinear activation function and encode them onto optical pulses input into the next layer.

Deep neural networks (DNNs) have become very popular in many areas, especially classification and prediction. However, as the number of neurons in the DNN increases to solve more complex problems, the DNN becomes limited by the latency and power consumption of existing hardware. A scalable, ultra-low latency photonic tensor processor can compute DNN layer outputs in a single shot. The processor includes free-space optics that perform passive optical copying and distribution of an input vector and integrated optoelectronics that implement passive weighting and the nonlinearity. An example of this processor classified the MNIST handwritten digit dataset (with an accuracy of 94%, which is close to the 96% ground truth accuracy). The processor can be scaled to perform near-exascale computing before hitting its fundamental throughput limit, which is set by the maximum optical bandwidth before significant loss of classification accuracy (determined experimentally).