Scalable extensible neural network system and methods

Patent metadata

Field	Value
Publication number	US-11488011-B2
Application number	US-201916352669-A
Country	US
Kind code	B2
Filing date	Mar 13, 2019
Priority date	Mar 13, 2019
Publication date	Nov 1, 2022
Grant date	Nov 1, 2022

Abstract

A neural network system, involving a neural network, the neural network configured to: map sensor output to a Level 1 input; learn to fuse the time slices for one class, learning comprising taking and feeding a random assignment of inputs from each time slice into a threshold function for another two-dimensional array; learn to reject class bias for completing network training; use cycles for class recognition, and fuse segments for intelligent information dominance and a magnetic headwear apparatus operably coupled with the neural network.

First claim

What is claimed:

1. A neural network system, comprising:
a neural network, comprising a two-dimensional array of nodes, where each node represents an artificial neuron in the neural network, configured to achieve intelligent information dominance through tractable deep learning, comprising:
a processor and non-transient memory device storing a set of executable instructions configured to:
map sensor output at a first level wherein at the first level there is a sensor output for each node of the neural network;
randomly sampling consecutive sensor outputs of the plurality of time slices to:
fuse a plurality of time slices to determine a class,
reject class bias in completing training;
detect the number of cycles to determine class recognition; and
fuse segments characterizing the two-dimensional arrays as well as in their associated weight granularities for use in hierarchical classification and recognition, and

a magnetic headwear apparatus operably coupled with the neural network.

2. The system of claim 1 , wherein the magnetic headwear apparatus comprises:
a headwear portion; and
an array of superconducting quantum interfere devices (“SQUIDs”) disposed in relation to the headwear portion, wherein the array of SQUIDS are configured to detect a magnetic signal in a range of 10 −9 teslas to 10 −6 teslas.

3. The system of claim 2 , wherein the headwear portion is configured to dispose the array of SQUIDs in close proximity to a scalp in a range of 0.1 in to 0.13 in.

4. The system of claim 2 , wherein the magnetic headwear apparatus further comprises a cryogenic feature for facilitating cooling the array of SQUIDs.

5. The system of claim 3 , wherein the array of SQUIDs is configured to:
detect magnetic signals emanating from a brain through the scalp, and
deliver the detected magnetic signals to the neural network.

6. The system of claim 1 , wherein the neural network is linear in its scalability.

7. The system of claim 1 , wherein the neural network is configured to at least one of learn and deep-learn fusing a plurality of spatial-temporal signatures.

8. The system of claim 5 , wherein the neural network is configured to detect and interpret the magnetic signals from the brain.

9. The system of claim 5 , wherein the neural network is configured to directly translate the magnetic signals from the brain.

10. The system of claim 1 , wherein the neural network is configured to provide a context in relation to diagnostic information.

11. A method of fabricating a neural network system, comprising:
providing a tractable neural network, the tractable neural network configured to:
map sensor output at a first level wherein at the first level there is a sensor output for each node of the neural network;
randomly sampling consecutive sensor outputs of the plurality of time slices to:
fuse a plurality of time slices to determine a class,
reject class bias in completing training;
detect the number of cycles to determine class recognition; and
fuse segments characterizing the two-dimensional arrays as well as in their associated weight granularities for use in hierarchical classification and recognition; and

providing a magnetic headwear apparatus operably coupled with the neural network.

12. The method of claim 11 , wherein the magnetic headwear apparatus comprises:
a headwear portion; and
an array of superconducting quantum interfere devices (“SQUIDs”) disposed in relation to the headwear portion.

13. The method of claim 12 , wherein the headwear portion is configured to dispose the array of SQUIDs in close proximity to a scalp in a range of approximately 0.1 in to approximately 0.13 in.

14. The method of claim 12 , wherein the magnetic headwear apparatus further comprises a cryogenic feature for facilitating cooling the array of SQUIDs.

15. The method of claim 13 , wherein the array of SQUIDs is configured to:
detect magnetic signals emanating from a brain through the scalp, and
deliver the detected magnetic signals to the neural network.

16. The method of claim 11 , wherein the neural network is linear in its scalability.

17. The method of claim 11 , wherein the neural network is configured to at least one of learn and deep-learn fusing a plurality of spatial-temporal signatures.

18. The method of claim 15 , wherein the neural network is configured to detect and interpret the magnetic signals from the brain.

19. The method of claim 15 ,
wherein the neural network is configured to directly translate the magnetic signals from the brain, and
wherein the neural network is configured to provide a context in relation to diagnostic information.

20. A method of improving deep learning by way of a tractable neural network system, comprising:
providing the tractable neural network system, comprising:
providing a neural network, the neural network configured to:
map sensor output at a first level wherein at the first level there is a sensor output for each node of the neural network;
randomly sampling consecutive sensor outputs of the plurality of time slices to:
fuse a plurality of time slices to determine a class,
reject class bias in completing training;
detect the number of cycles to determine class recognition; and

fuse segments characterizing the two-dimensional arrays as well as in their associated weight granularities for use in hierarchical classification and recognition; and

providing a magnetic headwear apparatus operably coupled with the neural network;

map sensor output at a first level wherein at the first level there is a sensor output for each node of the neural network;
randomly sampling consecutive sensor outputs of the plurality of time slices to:
fuse a plurality of time slices to determine a class,
reject class bias in completing training;
detect the number of cycles to determine class recognition; and

fuse segments characterizing the two-dimensional arrays as well as in their associated weight granularities for use in hierarchical classification and recognition.

Assignees

• Spawar Systems Ct Pacific
• Us Navy

Inventors

• Rubin Stuart H

Classifications

• A61B5/7267 Primary

  involving training the classification device · CPC title

• G06V10/82

  using neural networks · CPC title

• G06V10/776

  Validation; Performance evaluation · CPC title

• G06V10/764

  using classification, e.g. of video objects · CPC title

• G06N3/08 Primary

  Learning methods · CPC title