IP Library Granted Patent US 12,061,978
Granted Patent B2
US 12,061,978 · App. 17/041,333 · Granted Aug 13, 2024

Optical neural network unit and optical neural network configuration

Inventors: Zeev Zalevsky (Rosh HaAyin, IL); Eyal Cohen (Gedera, IL)
Assignee: BAR ILAN UNIVERSITY
G06N3/067G02F1/353G06F18/214G06F18/217G06N3/08G06V10/82G06V30/18057G06V30/18105G06V30/199G06V30/10
View Patent ↗
Loading inventors, assignments & file history…
Monitor This Case
Get email alerts when status or documents change.
Order Certified Copies
Most orders are placed with the USPTO same day — all within 24 business hours.
Order via The Patent Place →
Pre-filled with this patent's details
Quick Facts
Patent No.
US 12,061,978
App. No.
17/041,333
Granted
Aug 13, 2024
Kind
B2
Abstract

An artificial neuron unit and neural network for processing of input light are described. The artificial neuron unit comprises a modal mixing unit, such as multimode optical fiber, configured for receiving input light and applying selected mixing to light components of two or more modes within the input light and for providing exit light, and a filtering unit configured for applying preselected filter onto said exit light for selecting one or more modes of the exit light thereby providing output light of the artificial neuron unit.

Claims (45)

1. An artificial neuron unit for processing of input light, the artificial neuron unit comprising:

a modal mixing unit configured for receiving input light and applying selected mixing to light components of two or more modes within the input light providing exit light, said modal mixing unit being configured as a multimode optical fiber (MMF) having a first end and a second end, and being configured for receiving the input light at the first end, enabling propagation of the input light through the MMF while mixing spatial modes of the input light propagating in respective velocities within the MMF to yield an exit light, and for outputting the exit light at the second end;

an input optical arrangement, configured for coupling the input light into the first end of the MMF;

a filtering unit configured for applying preselected filter onto said exit light for selecting one or more modes of the exit light, thereby providing output light of the artificial neuron unit, said filtering unit being configured as a spatial light modulator (SLM), configured for imposing a selected spatially varying modulation on the exit light to yield an output light; and

a control unit configured and operable for operating said SLM and for determining spatial light modulation applied thereby, said control unit being configured for selecting spatial modulation to output light in accordance with training process of a neural processing network comprising said unit.

2. The artificial neuron unit of claim 1 , wherein said model mixing unit is configured for mixing two or more modes selected by at least one of the following: polarization orientation modes, wavelength ranges, spatial modes within a selected region, or spatial modes within two or more cores of the model mixing unit.

3. The artificial neuron unit of claim 1 , wherein said modal mixing unit is configured for applying linear mixing thereby providing said exit light being weighted linear combination of two or more modes of the input light.

4. The artificial neuron unit of claim 1 , further comprising an output optical arrangement configured for interacting with the output light.

5. The artificial neuron unit of claim 1 , further comprising a feedback route configured for receiving at least a portion of the exit light at said second end of the MMF and directing light components of said at least a portion of exit light toward said first end of the MMF for mixing said light components with at least a portion of input light, said feedback route being associated with an output port being associated with said spatial light modulator.

6. The artificial neuron unit of claim 5 , wherein said output port of the feedback route being an auxiliary output port configured for outputting light signals associated with said mixing of said light components collected via the feedback route with at least a portion of input light.

7. The artificial neuron unit of claim 5 , wherein said output port of the feedback route is configured for providing output associated with at least a portion of the output light.

8. The artificial neuron unit of claim 5 , wherein said feedback route comprises gain unit and is configured for transmitting least a portion of the exit light through said gain unit for increasing intensity thereof.

9. The artificial neuron unit of claim 1 , wherein said artificial neuron unit is located at input port of a neural network structure and configured for applying selected pre-processing to light signals provided to a neural network processing structure.

10. An artificial neuron network, comprising:

one or more artificial neuron units for processing of input light, the artificial neuron unit comprising a modal mixing unit configured for receiving input light and applying selected mixing to light components of two or more modes within the input light providing exit light, and a filtering unit configured for applying preselected filter onto said exit light for selecting one or more modes of the exit light thereby providing output light of the artificial neuron unit;

wherein said one or more artificial neuron units are arranged in one or more neuron layers along a path of propagation of an optical signal, such that the optical signal is configured to propagate through the one or more artificial neuron units between input ports of artificial neuron units of an input layer to output ports of artificial neuron units of a output layer providing output signal of said network.

11. The artificial neuron network of claim 10 , further comprising one or more feedback route configured for receiving at least one portion of output light from at least one output port of an artificial neuron unit of said output layer and directing at least a portion of the output light for mixing with at least a portion of input light directed at artificial neuron units of the input layer, and for outputting at least a portion of the mixed light.

12. The artificial neuron network of claim 11 wherein the feedback route comprises:

a feedback unit, configured for receiving the output light; and

an X-coupler having a first and a second input end and a first and a second output end, and configured for receiving the at least one portion of output light from the feedback unit via the first input end, receiving the input light in via the second input end, mixing the input light and the output light to yield the mixed light, directing the at least first portion of the mixed light into the MMF's first end, and outputting the at least a second portion of the mixed light.

13. The artificial neuron network of claim 12 , wherein said artificial neuron network further comprises an all-optical light modulator located at second output end of the X-coupler, said all-optical light modulator being configured as a liquid crystal valve.

14. The artificial neuron network of claim 12 , wherein said artificial neuron network further comprises a nonlinear light modulator located at second output end of the X-coupler and configured for applying one or more nonlinear interactions to light components passing therethrough, said one or more nonlinear interactions comprises at least one of second harmonic generation, sum frequency generation, difference frequency generation.

15. The artificial neuron network of claim 12 , wherein the feedback route comprises:

a first semi-transparent mirror located near or at the second end of the MMF; and

a second semi-transparent mirror located near or at the first end of the MMF;

wherein:

the first mirror is confirmed for reflecting the at least one portion of the output light back into the second MMF, such that the at least one portion of the output light enters the MMF via the second end and exit the MMF via the first end;

the second mirror is configured for reflecting the at least one portion of the output light back into the MMF via the first end, while transmitting at least one portion of the input light into the MMF via the first end, such that the at least one portion of the input light and the at least one portion mix in the MMF yielding mixed light;

the first mirror is configured for transmitting at least a portion of the mixed light, such that the artificial neuron network is configured for outputting the at least one portion of the mixed light.

16. The artificial neuron network of claim 12 , wherein the feedback route comprises one or more optical fibers configured for directing said at least a portion of the output light for mixing with said input light.

17. The artificial neuron network of claim 16 , wherein said one or more optical fibers comprising one or more of: single-core fiber, multi-core fiber, and a bundle of optical fibers.

18. An artificial neuron unit for processing of input light, the artificial neuron unit comprising:

a modal mixing unit configured for receiving input light and applying selected mixing to light components of two or more modes within the input light providing exit light, said modal mixing unit being configured as a multimode optical fiber (MMF) having a first end and a second end, and being configured for receiving the input light at the first end, enabling propagation of the input light through the MMF while mixing spatial modes of the input light propagating in respective velocities within the MMF to yield an exit light, and for outputting the exit light at the second end,

an input optical arrangement, configured for coupling the input light into the first end of the MMF;

a filtering unit configured for applying preselected filter onto said exit light for selecting one or more modes of the exit light thereby providing output light of the artificial neuron unit, said filtering unit being configured as a spatial light modulator (SLM), configured for imposing a selected spatially varying modulation on the exit light to yield an output light; and

a feedback route configured for receiving at least a portion of the exit light at said second end of the MMF and directing light components of said at least a portion of exit light toward said first end of the MMF for mixing said light components with at least a portion of input light, said feedback route being associated with an output port being associated with said spatial light modulator.

19. The artificial neuron unit of claim 18 , further comprising a control unit configured and operable for operating said spatial light modulator (SLM) and for determining spatial light modulation applied thereby.

20. The artificial neuron unit of claim 19 , wherein said control unit is configured for selecting spatial modulation to output light in accordance with training process of a neural processing network comprising said unit.

21. The artificial neuron unit of claim 18 , wherein said model mixing unit is configured for mixing two or more modes being selected by at least one of the following: polarization orientation modes, wavelength ranges, spatial modes within a selected region and spatial modes within two or more cores of the model mixing unit.

22. The artificial neuron unit of claim 18 , wherein said modal mixing unit is configured for applying linear mixing thereby providing said exit light being weighted linear combination of two or more modes of the input light.

23. The artificial neuron unit of claim 18 , further comprising an output optical arrangement configured for interacting with the output light.

24. The artificial neuron unit of claim 18 , wherein said output port of the feedback route is an auxiliary output port configured for outputting light signals associated with said mixing of said light components collected via the feedback route with at least a portion of input light.

25. The artificial neuron unit of claim 18 , wherein said output port of the feedback route is configured for providing output associated with at least a portion of the output light.

26. The artificial neuron unit of claim 18 , wherein said feedback route comprises gain unit and is configured for transmitting least a portion of the exit light through said gain unit for increasing intensity thereof.

27. The artificial neuron unit of claim 18 , configured for locating at input port of a neural network structure and configured for applying selected pre-processing to light signals provided to a neural network processing structure.

Assignments (2)
ASSIGNMENT OF ASSIGNOR'S INTEREST Recorded Sep 24, 2020
From: ZALEVSKY, ZEEV; COHEN, EYAL
To: BAR ILAN UNIVERSITY
Reel/Frame 053877/0177 →
ASSIGNMENT OF ASSIGNOR'S INTEREST Recorded Sep 24, 2020
From: ZALEVSKY, ZEEV; COHEN, EYAL
To: BAR ILAN UNIVERSITY
Reel/Frame 053877/0228 →
Continuity (2)
Provisional Application 62648538 · Mar 27, 2018
Related Publication 20210027154A1 · Jan 28, 2021
Cited By (9)
US 12,353,988 US 12,564,047 US 12,568,809 US 12,615,087 US 12,674,948 US 12,674,949 US 12,690,438 US 12,699,240 US 12,706,680