Abstract: Convolutional neural networks have, over the last decade, risen to state-of-the-art for computer vision tasks such as image classification. Oftentimes these are implemented on specialized digital processors, which target high throughput operation. This is efficient when analyzing batches of images loaded from memory but is lacking when analyzing freshly-captured images on a sensor. Latency is an issue since image acquisition alone can take as long as the processing. Second, streams of images captured from sensors can result in highly redundant processing if only specific signatures needs to be recognized, leading to a high amount of wasted power. Disclosed is a way to perform convolutional processing on the image plane where the image is captured, which allows the device to only complete readout and processing when required. This reduces latency and power consumption, enabling new applications such as augmented reality and edge processing that are not possible with current technology.

Abstract: A reconfigurable thin-film photonic filter weight bank comprises at least one photodetector and at least one optical filtering device comprising a pair of reflective thin film stacks with an interstitial medium cavity therebetween forming an optical cavity. Operation of a bank occurs when the reflective film is more selectively reflective to a range of frequencies and is more translucent to frequencies outside the range. The bank can reconfigurably weight signals by varying the cavity sizes or complex indices of refraction, with one photodetector integrating the weighted signals. Detectors can also be embedded inside the individual cavities to output unweighted, but demultiplexed, electrical signals. A neuromorphic opto-electronic system comprises a plurality of interconnected artificial optical neurons, each including at least one thin film neuromorphic opto-electronic device with a reconfigurable thin-film photonic filter weight bank. Related methods are also disclosed.

Abstract: A hybrid neuromorphic computing device is provided, in which artificial neurons include light-emitting devices that provide weighted sums of inputs as light output. The output is detected by a photodetector and converted to an electrical output. Each neuron may receive output from one or more other neurons as initial input. Interconnects between neurons may be optical, electrical, or a combination thereof. The neurons also may provide imaging sensor and/or display capabilities.

Abstract: A processing device comprises a plurality of artificial neurons comprising first and second artificial neurons, at least one optical connection between an optical output of the first artificial neuron and an optical input of the second artificial neuron, the optical input comprising a photosensitive element, at least one weighting element connected to the optical input of the second artificial neuron, configured to modify a gain of the optical input of the second artificial neuron, and at least one nonlinear element having an input and connected to the optical output of the first artificial neuron, configured to activate the optical output when a signal received at the input of the at least one nonlinear element rises above a threshold, wherein the optical output of the first artificial neuron comprises at least one thin film light-generating component. An LED neuron is also described.