Physical neural network design incorporating nanotechnology

• US 20030177450A1
• Filed: 03/12/2002
• Published: 09/18/2003
• Est. Priority Date: 03/12/2002
• Status: Active Grant

• Current Assignee: Knowm Tech LLC
• Original Assignee: Knowm Tech LLC
• Inventors: Nugent, Alex
• Granted Patent: US 6,889,216 B2
• Time in Patent Office: Days
• US Class Current: 716/1
• CPC Class Codes: B82Y 10/00 Nanotechnology for information processing, storage or transmission, e.g. quantum computing or single electron logic
• CPC Class Codes: G06N 3/063 using electronic means
• CPC Class Codes: Y10S 977/932 for electronic or optoelectronic application

Abstract

A physical neural network based on nanotechnology, including methods thereof. Such a physical neural network generally includes one or more neuron-like nodes, which are formed from a plurality of interconnected nanoconnections formed from nanoconductors. Each neuron-like node sums one or more input signals and generates one or more output signals based on a threshold associated with the input signal. The physical neural network also includes a connection network formed from the interconnected nanoconnections, such that the interconnected nanoconnections used thereof by one or more of the neuron-like nodes are strengthened or weakened according to an application of an electric field.

Claims

1. A physical neural network based on nanotechnology, said physical neural network comprising:

at least one neuron-like node that sums at least one input signal and generates at least one output signal based on a threshold associated with said at least one input signal; and

at least one connection network associated with said at least one neuron-like node wherein said at least one connection network comprises a plurality of interconnected nanoconnections, such that each nanoconnection of said plurality of interconnected nanoconnections is strengthened or weakened according to an application of an electric field, thereby providing.

2. The physical neural network of claim 1 wherein said at least one output signal comprises a non-linear output signal based on said threshold.

3. The physical network of claim 1 wherein said at least one output signal comprises a linear output signal based on said threshold.

4. The physical neural network of claim 1 wherein said threshold comprises a threshold below which said at least one output signal is not generated and above which said at least one output signal is generated.

5. The physical neural network of claim 1 wherein said at least one connection network comprises:

a number of layers of said nanoconnections;

wherein said number of layers is equal to a number of desired outputs from said at least one connection network; and

wherein said nanoconnections are formed without influence by disturbances resulting from other nanoconnections thereof.

6. The physical neural network of claim 1 wherein at least one nanoconnection of said plurality of interconnected nanoconnections comprises an electrically conducting material.

7. The physical neural network of claim 6 wherein electrically conducting material is chosen such that a dipole is induced in said electrically conducting material in the presence of an electric field.

8. The physical neural network of claim 6 wherein said electrically conducting material comprises a chemically induced permanent dipole.

9. The physical neural network of claim 1 wherein said at least one nanoconnection comprises at least one nanoconductor.

10. The physical neural network of claim 9 wherein said at least one connection network comprises:

at least one connection network structure having a connection gap formed therein;

a solution located within said connection gap;

wherein said solution comprises a solvent and said at least one nanoconductor; and

wherein an electric field applied across said connection gap to permit an alignment of at least one nanoconductor within said connection gap.

11. The physical neural network of claim 10 wherein said at least one nanoconductor is suspended in said solvent.

12. The physical neural network of claim 10 wherein said at least one nanoconductor is located at a bottom of said connection gap.

13. The physical neural network of claim 10 wherein an electrical conductance of said solution is less than an electrical conductance of said at least one nanoconductor within said solution.

14. The physical neural network of claim 10 wherein a viscosity of said solution permits said at least one nanoconductor to move when said electric field is applied across said connection gap.

15. The physical neural network of claim 10 wherein said at least one nanoconductor experiences an increased alignment in accordance with an increase in said electric field applied across said connection gap.

16. The physical neural network of claim 15 wherein nanoconnections of said at least one neuron-like node that are utilized most frequently by said at least one neuron-like node become stronger with each use thereof.

17. The physical neural network of claim 16 wherein said nanoconnections that are utilized least frequently become increasingly weak and eventually become unaligned.

18. The physical neural network of claim 16 wherein said at least one nanoconnection comprises a resistance, which is raised or lowered by a selective activation of said at least one nanoconnection.

19. The physical neural network of claim 9 wherein said at least one nanoconductor comprises a nanowire.

20. The physical neural network of claim 9 wherein said at least one nanoconductor comprises a nanotube.

21. The physical neural network of claim 9 wherein said at least one nanoconductor comprises a plurality of nanoparticles.

22. The physical neural network of claim 1 wherein at least one nanoconnection of said at least one connection network comprises a negative connection associated with said at least one neuron-like node.

23. The physical neural network of claim 22 wherein said at least one connection network comprises:

a number of layers of nanoconnections;

wherein said number of layers is equal to a number of desired outputs from said at least one connection network;

wherein said number of layers is equal to twice a number of desired outputs form said at least one connection network, if said nanoconnections comprise negative connections thereof; and

wherein said nanoconnections are formed without influence by disturbances resulting from other nanoconnections thereof.

24. A physical neural network apparatus based on nanotechnology, said physical neural apparatus comprising:

at least one connection network associated with at least one neuron-like node wherein said at least one connection network comprises a plurality of interconnected nanoconductors, such that each nanoconductor of said plurality of interconnected nanoconductors is strengthened or weakened according to an application of an electric field;

wherein each nanoconductor of said plurality of interconnected nanoconductors experiences an increase in alignment in accordance with an increase in said electric field;

wherein nanoconductors of said plurality of interconnected nanoconductors that are utilized most frequently by said at least one neuron-like node become stronger with each use thereof; and

wherein nanoconductors of said plurality of interconnected nanoconductors that are utilized least frequently become increasingly weak and eventually become unaligned.

25. The physical neural network apparatus of claim 24 wherein said at least one neuron-like node sums at least one input signal and generates at least one output signal based on a threshold associated with said at least one input signal.

26. The physical neural network of claim 24 wherein said at least one connection network comprises a plurality of layers, wherein a number of layers of said plurality of layers is equal to a desired number of outputs from said at least one connection network.

27. A method for assembling a physical neural network based on nanotechnology, said method comprising the steps of:

forming at least one neuron-like node that sums at least one input signal and generates at least one output signal based on a threshold associated with said at least one input signal; and

configuring at least one connection network associated with said at least one neuron-like node wherein said at least one connection network comprises a plurality of interconnected nanoconnections, such that each nanoconnection of said plurality of interconnected nanoconnections is strengthened or weakened according to an application of an electric field, thereby providing.

28. The method of claim 27 wherein said at least one output signal comprises a non-linear output signal based on said threshold.

29. The method of claim 27 wherein said at least one output signal comprises a linear output signal based on said threshold.

30. The method of claim 27 wherein said threshold comprises a threshold below which said at least one output signal is not generated and above which said at least one output signal is generated.

31. The method of claim 27 further comprising the steps of:

configuring said at least one connection network to comprise a number of layers of said nanoconnections, wherein said number of layers is equal to a number of desired outputs from said at least one connection network; and

forming said nanoconnections without influence by disturbances resulting from other nanoconnections thereof.

32. The method of claim 27 further comprising the step of:

forming at least one nanoconnection of said plurality of interconnected nanoconnections from an electrically conducting material.

33. The method of claim 32 further comprising the step of:

configuring said electrically conducting material such that a dipole is induced in said electrically conducting material in the presence of an electric field.

34. The method of claim 32 further comprising the step of:

chemically inducing a permanent dipole within said electrically conducting material.

35. The method of claim 27 wherein said at least one nanoconnection comprises at least one nanoconductor.

36. The method of claim 35 further comprising the steps of:

forming a connection gap from at least one connection network structure associated with said at least one connection network;

locating a solution within said connection gap;

configuring said solution to comprises a solvent and said at least one nanoconductor; and

applying an electric field across said connection gap to permit an alignment of at least one nanoconductor within said connection gap.

37. The method of claim 36 further comprising the step of:

suspending said at least one nanoconductor in said solvent.

38. The method of claim 36 further comprising the step of:

locating said at least one nanoconductor at a bottom of said connection gap.

39. The method of claim 36 further comprising the step of:

configuring said solution such that an electrical conductance of said solution is less than an electrical conductance of said at least one nanoconductor within said solution.

40. The method of claim 36 further comprising the step of:

configuring said solution to comprise a viscosity that permits said at least one nanoconductor to move when said electric field is applied across said connection gap.

41. The method of claim 36 further comprising the step of:

increasing an application of said electric field across said connection gap to thereby permit said at least one nanoconductor to experience an increased alignment in accordance with an increase in said electric field applied across said connection gap.

42. The method of claim 41 wherein nanoconnections of said at least one neuron-like node that are utilized most frequently by said at least one neuron-like node become stronger with each use thereof.

43. The method of claim 42 wherein said nanoconnections that are utilized least frequently become increasingly weak and eventually become unaligned.

44. The method of claim 42 wherein said at least one nanoconnection comprises a resistance, which is raised or lowered by a selective activation of said at least one nanoconnection.

45. The method of claim 35 further comprising the step of:

configuring said at least one nanoconductor to comprise a nanowire.

46. The method of claim 35 further comprising the step of:

configuring said at least one nanoconductor to comprise a nanotube.

47. The method of claim 35 further comprising the step of:

configuring said at least one nanoconductor to comprise a plurality of nanoparticles.

48. The method of claim 27 wherein at least one nanoconnection of said at least one connection network comprises a negative connection associated with said at least one neuron-like node.

49. The method of claim 48 further comprising the step of:

configuring said at least one connection network to comprise a number of layers of nanoconnections, wherein said number of layers is equal to a number of desired outputs from said at least one connection network;

wherein said number of layers is equal to twice a number of desired outputs form said at least one connection network, if said nanoconnections comprise negative connections thereof; and

wherein said nanoconnections are formed without influence by disturbances resulting from other nanoconnections thereof.

50. A method for assembling a physical neural network apparatus based on nanotechnology, said method comprising the steps of:

forming at least one connection network associated with at least one neuron-like node wherein said at least one connection network comprises a plurality of interconnected nanoconductors, such that each nanoconductor of said plurality of interconnected nanoconductors is strengthened or weakened according to an application of an electric field;

wherein each nanoconductor of said plurality of interconnected nanoconductors experiences an increase in alignment in accordance with an increase in said electric field;

wherein nanoconductors of said plurality of interconnected nanoconductors that are utilized most frequently by said at least one neuron-like node become stronger with each use thereof; and

wherein nanoconductors of said plurality of interconnected nanoconductors that are utilized least frequently become increasingly weak and eventually become unaligned.

51. The method apparatus of claim 50 further comprising the steps of:

summing at least one input signal via said at least one neuron-like node; and

generating at least one output signal based on a threshold associated with said at least one input signal.

52. The method of claim 50 further comprising the step of:

configuring said at least one connection network to comprise a plurality of layers, wherein a number of layers of said plurality of layers is equal to a desired number of outputs from said at least one connection network.