Hybrid neuromorphic computing display
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, allowing for high degrees of fan-out and fan-in, including true broadcast-to-all functionality.
1. A processing device comprising:
a plurality of communicatively-coupled thin film artificial neurons, each artificial neuron comprising:
one or more thin film light-generating components; and
a photodetector configured to sum outputs of the one or more light generating components to an output of the photodetector;
wherein each output of each of the one or more thin film light-generating components is determined by one or more inputs to the each light-generating component.
2. The processing device of claim 1 , wherein, for a first plurality of the plurality of artificial neurons, the photodetector output of each neuron of the first plurality of neurons is provided as one of the one or more inputs of at least one other neuron of the first plurality of neurons.
3. The processing device of claim 2 , wherein weighting factors are applied to the inputs of the one or more light-generating components.
4. The processing device of claim 3 , wherein the weighting factors are obtained from a source external to the plurality of neurons.
5. The processing device of claim 3 , wherein the weighting factors are obtained from a source stored in memory within each of the plurality of neurons.
6. The processing device of claim 1 , wherein an output of the photodetector is provided to a threshold device that provides an output based upon the output of the photodetector, and the output of the threshold device is provided as an input to one or more of the plurality of neurons.
7. The processing device of claim 1 , wherein each artificial neuron comprises a plurality of photodetectors.
8. The processing device of claim 1 , wherein the output of the photodetector is provided to a threshold device, the output of which is greater than zero only if the sum of the outputs of photodetectors providing outputs to the threshold device exceeds a threshold value.
9. The processing device of claim 8 , wherein, if the sum of the outputs of the photodetectors exceeds a threshold value, the output of the threshold device is in the form of a spike.
10. The processing device of claim 1 where the input signals are in the form of a spike.
11. The processing device of claim 1 , wherein the processing device is also a display panel.
12. The processing device of claim 1 , wherein each of the one or more light-generating components comprises an organic light emitting diode (OLED).
13. The processing device of claim 1 , wherein the photodetector comprises a thin film photodetector disposed within the device such that light emitted by the light-generating component is incident on the thin film photodetector.
14. The processing device of claim 1 , wherein each of the plurality of artificial neurons is electrically or optically connected to each other of the plurality of artificial neurons such that an output of each threshold device from each neuron is available as an input to one or more other artificial neurons of the plurality of neurons.
15. The processing device of claim 12 , wherein the OLED in each artificial neuron of the plurality of artificial neurons is driven by a plurality of row and column lines.
16. The processing device of claim 1 , wherein the device is operable as an imaging device.
17. The processing device of claim 16 , wherein the device comprises a plurality of pixels and each of the plurality of artificial neurons comprises one or more of the plurality of pixels.
18. The processing device of claim 16 , wherein the device is further operable as a display panel.
19. The processing device of claim 18 , wherein a portion of an active area of the display panel is reserved for use as artificial neurons or imaging pixels.
20. The processing device of claim 1 , wherein each artificial neuron comprises a plurality of light generating components.
21. The processing device of claim 1 , wherein each artificial neuron of the plurality of artificial neurons is electrically connected to at least one other artificial neuron of the plurality of artificial neurons.
22. The processing device of claim 1 , wherein each artificial neuron of the plurality of artificial neurons is optically connected to at least one other artificial neuron of the plurality of artificial neurons.
23. A method of operating an artificial neuron array comprising an array of neurons arranged in a plurality of rows and a plurality of columns, the method comprising:
selecting a first row of the plurality of rows of neurons;
receiving weighting factor pairs for each neuron in the first row;
outputting, by each artificial neuron in the first row, a sum of weighted input values, the weighted input values being represented by light emitted by a light-emitting device of the each artificial neuron based upon the weighting factor pairs for the each artificial neuron and one or more input values received by the each artificial neuron;
detecting, by a photodiode in each neuron, the light emitted by the light-emitting device in the each neuron; and
for each neuron, based upon the detected light emitted by the light-emitting device in the each neuron and a threshold value, outputting a value of 0 or 1.