MEMRISTIVE NANOFIBER NEURAL NETWORKS

• US 20170098156A1
• Filed: 12/19/2016
• Published: 04/06/2017
• Est. Priority Date: 06/19/2014
• Status: Active Grant

• Current Assignee: University of Florida Research Foundation, Incorporated (State University System of Florida)
• Original Assignee: University of Florida Research Foundation, Incorporated (State University System of Florida)
• Inventors: Kendall, Jack, Nino, Juan Claudio
• Granted Patent: US 10,198,691 B2
• Time in Patent Office: Days
• CPC Class Codes: G06N 3/044 Recurrent networks, e.g. Hopfield networks
• CPC Class Codes: G06N 3/049 Temporal neural networks, e.g. delay elements, oscillating neurons or pulsed inputs
• CPC Class Codes: G06N 3/065 Analogue means
• CPC Class Codes: G06N 3/08 Learning methods
• CPC Class Codes: G11C 13/0007 comprising metal oxide memory material, e.g. perovskites

Abstract

Disclosed are various embodiments of memristive neural networks comprising neural nodes. Memristive nanofibers are used to form artificial synapses in the neural networks. Each memristive nanofiber may couple one or more neural nodes to one or more other neural nodes. In one case, a memristive neural network includes a first neural node, a second neural node, and a memristive fiber that couples the first neural node to the second neural node. The memristive fiber comprises a conductive core and a memristive shell, where the conductive core forms a communications path between the first neural node and the second neural node and the memristive shell forms a memristor synapse between the first neural node and the second neural node.

Claims

1. A memristive neural network, comprising:

a first neural node;

a second neural node; and

a memristive fiber that couples the first neural node to the second neural node, wherein the memristive fiber comprises a conductive core and a memristive shell, wherein the conductive core forms a communications path between the first neural node and the second neural node, wherein the memristive shell forms a memristor synapse between the first neural node and the second neural node.

2. The memristive neural network of claim 1, wherein the first neural node and the second neural node are among a plurality of neural nodes in a neural node array.

3. The memristive neural network of claim 2, wherein each of the neural nodes comprises a respective Leaky Integrate-and-Fire (LIF) Complimentary Metal-Oxide-Semiconductor (CMOS) neural circuit.

4. The memristive neural network of claim 2, wherein the memristive fiber is among a plurality of memristive fibers in a memristive fiber network, wherein at least a subset of the plurality of memristive fibers in the memristive fiber network are randomly coupled to at least a subset of the plurality of neural nodes in the neural node array.

5. The memristive neural network of claim 4, wherein the memristive fiber network comprises at least one recurrent connection.

6. The memristive neural network of claim 4, wherein the memristive fiber network comprises at least one inhibitory output for at least one of the plurality of neural nodes.

7. The memristive neural network of claim 4, wherein the memristive fiber network comprises a plurality of memristive fiber layers.

8. The memristive neural network of claim 7, wherein the memristive fiber network comprises at least one connection that facilitates a transmission of at least one signal between multiple ones of the plurality of memristive fiber layers.

9. The memristive neural network of claim 1, wherein the memristive fiber comprises:

a first electrode that couples to the first neural node; and

a second electrode that couples to the second neural node.

10. The memristive neural network of claim 1, wherein a Liquid State Machine (LSM) is modeled by at least the first neural node, the second neural node, and the memristive fiber.

11. The memristive neural network of claim 1, wherein the memristive fiber is electrospun to facilitate coupling between the first neural node and the second neural node.

12. The memristive neural network of claim 11, wherein the memristive shell comprises TiO₂ and the conductive core is doped with TiO_2-x.

13. The memristive neural network of claim 1, wherein the conductive core comprises TiO_2-x.

14. The memristive neural network of claim 1, wherein the first neural node outputs at least one signal in response to receiving at least one input signal.

15. The memristive neural network of claim 1, wherein the memristive fiber is one of a plurality of memristive fibers, and wherein individual ones of the plurality of memristive fibers form randomized connections between the first neural node and the second neural node.

16. A method, comprising:

providing a first neural node;

providing a second neural node;

coupling the first neural node to the second neural node using at least a memristive fiber, wherein the memristive fiber comprises a conductive core and a memristive shell, wherein the conductive core forms a communications path between the first neural node and the second neural node, wherein the memristive shell forms a memristor synapse between the first neural node and the second neural node.

17. The method of claim 16, wherein the first neural node and the second neural node are among a plurality of neural nodes in a neural node array.

18. The method of claim 16, wherein the first neural node and the second neural node comprises a respective Leaky Integrate-and-Fire (LIF) Complimentary Metal-Oxide-Semiconductor (CMOS) neural circuit.

19. The method of claim 18, wherein the conductive core comprises TiO_2-x.

20. The method of claim 16, further comprising randomly coupling a plurality of memristive fibers to a plurality of neural nodes