Machine generated content naming in an information centric network

• US 11,637,772 B2
• Filed: 06/28/2019
• Issued: 04/25/2023
• Est. Priority Date: 06/28/2019
• Status: Active Grant

• Current Assignee: Intel Corporation
• Original Assignee: Intel Corporation
• Inventors: Nallampatti Ekambaram, Venkatesan, Jha, Satish Chandra, Smith, Ned M., Alam, S. M. Iftekharul, Ramirez Loaiza, Maria, Zhang, Yi, Arrobo Vidal, Gabriel
• Primary Examiner(s): Huang, Miranda M
• Assistant Examiner(s): Bostwick, Sidney Vincent
• Application Number:
  US16/457,480
• Publication Number:
  US 20190327171A1
• Time in Patent Office: 1,397 Days
• CPC Class Codes: G06N 20/10 using kernel methods, e.g. support vector machines [SVM]
• CPC Class Codes: G06N 3/04 Architecture, e.g. interconnection topology
• 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 3/063 using electronic means
• CPC Class Codes: G06N 3/08 Learning methods
• CPC Class Codes: G06N 3/088 Non-supervised learning, e.g. competitive learning
• CPC Class Codes: G06N 7/01 Probabilistic graphical models, e.g. probabilistic networks
• CPC Class Codes: H04L 45/507 Label distribution
• CPC Class Codes: H04L 67/568 Storing data temporarily at an intermediate stage, e.g. caching
• CPC Class Codes: H04L 67/63 Routing a service request depending on the request content or context

Abstract

Systems and techniques for machine generation of content names in an information centric network (ICN) are described herein. For example, a node may obtain content. An inference engine may be invoked to produce a name for the content. Once the content is named, the node may respond to an interest packet that includes the name of the content. The response is a data packet that includes the content.

Claims

1. A device for machine generation of content names in an information centric network (ICN), the device comprising:

processing circuitry; and

a memory including instructions that, when executed by the processing circuitry, configure the processing circuitry to;

obtain content;

invoke a first inference engine with a spiking neural network to implement a string predictor that produces a name for the content, the spiking neural network implemented with spike timing dependent plasticity (STDP) learning to train the spiking neural network in an on-going manner to adapt the name over time, wherein the name generated by the string predictor does not conform to any defined standard; and

respond to an interest packet that includes the name with a data packet that includes the content, the name in the interest packet generated with a second inference engine at a subscriber device that created the interest packet, wherein the second inference engine uses another spiking neural network trained as the spiking neural network of the first inference engine to implement the string predictor, and wherein the device is a publisher device.

2. The device of claim 1, wherein the interest packet is received before the inference engine is invoked, and wherein an attribute of the interest packet was provided to the inference engine to produce the name.

3. The device of claim 2, wherein the attribute of the interest packet is at least one of an entity name that originated the interest packet or a location of an entity that originated the interest packet.

4. The device of claim 2, wherein the attribute is a content feature set.

5. The device of claim 1, wherein the spiking neural network is trained with a constraint that limits a form of the name.

16. The device of claim 5, wherein the constraint limits a size or content of part of the name to ensure more efficient routing in the ICN.

17. The device of claim 1, wherein the instructions configure the processing circuitry to:

receive a content feature of the content in addition to the content; and

provide the content feature set to the inference engine as part of the invocation of the inference engine.

22. The device of claim 1, wherein the device is part of a system that includes the different device, and wherein the different devices includes:

second processing circuitry; and

a second memory including second instructions that, when executed by the second processing circuitry, configure the second processing circuitry to;

invoke the copy of the inference engine to produce the name for the content;

create the interest packet with the name to request the content; and

transmit the interest packet.

6. A method for machine generation of content names in an information centric network (ICN), the method comprising:

obtaining content at a first device, wherein the first device is a publisher device;

invoking, by the first device, a first inference engine with a spiking neural network to implement a string predictor that produces a name for the content, the spiking neural network implemented with spike timing dependent plasticity (STDP) learning to train the spiking neural network in an on-going manner to adapt the name over time, wherein the name generated by the string predictor does not conform to any defined standard; and

responding, by the first device, to an interest packet that includes the name with a data packet that includes the content, the name in the interest packet generated by a second device, using a second inference engine, that created the interest packet, wherein the second inference engine uses another spiking neural network trained as the spiking neural network to implement the string predictor, and wherein the second device is a subscriber device.

7. The method of claim 6, wherein the interest packet is received before the inference engine is invoked, and wherein an attribute of the interest packet was provided to the inference engine to produce the name.

8. The method of claim 7, wherein the attribute of the interest packet is at least one of an entity name that originated the interest packet or a location of an entity that originated the interest packet.

9. The method of claim 7, wherein the attribute is a content feature set.

10. The method of claim 6, wherein the spiking neural network is trained with a constraint that limits a form of the name.

18. The method of claim 6, wherein the constraint limits a size or content of part of the name to ensure more efficient routing in an ICN.

19. The method of claim 6, comprising:

receiving a content feature of the content in addition to the content; and

providing the content feature set to the inference engine as part of the invocation of the inference engine.

23. The method of claim 6, comprising:

invoking, by the second device, the copy of the inference engine to produce the name for the content;

creating, by the second device, the interest packet with the name to request the content; and

transmitting, by the second device, the interest packet.

11. At least one non-transitory machine-readable medium including instructions for machine generation of content names in an information centric network (ICN), the instructions, when executed by processing circuitry of a first device, cause the processing circuitry to perform operations comprising:

obtaining content;

invoking a first inference engine with a spiking neural network to implement a string predictor that produces a name for the content, the spiking neural network implemented with spike timing dependent plasticity (STDP) learning to train the spiking neural network in an on-going manner to adapt the name over time, wherein the name generated by the string predictor does not conform to any defined standard; and

responding to an interest packet that includes the name with a data packet that includes the content, the name in the interest packet generated by a second device, using a second inference engine, that created the interest packet, wherein the second inference engine uses another spiking neural network trained as the spiking neural network to implement the string predictor, wherein the second device is a subscriber device, and wherein the first device is a publisher device.

12. The at least one non-transitory machine-readable medium of claim 11, wherein the interest packet is received before the inference engine is invoked, and wherein an attribute of the interest packet was provided to the inference engine to produce the name.

13. The at least one non-transitory machine-readable medium of claim 12, wherein the attribute of the interest packet is at least one of an entity name that originated the interest packet or a location of an entity that originated the interest packet.

14. The at least one non-transitory machine-readable medium of claim 12, wherein the attribute is a content feature set.

15. The at least one non-transitory machine-readable medium of claim 11, wherein the spiking neural network is trained with a constraint that limits a form of the name.

20. The at least one non-transitory machine-readable medium of claim 11, wherein the constraint limits a size or content of part of the name to ensure more efficient routing in an ICN.

21. The at least one non-transitory machine-readable medium of claim 11, wherein the operations comprise:

receiving a content feature of the content in addition to the content; and

providing the content feature set to the inference engine as part of the invocation of the inference engine.

24. The at least one non-transitory machine-readable medium of claim 11, wherein the first device is part of a system that includes the second device, and wherein the second device includes

a second at least one non-transitory machine readable medium including second instructions that, when executed by second processing circuitry, cause the second processing circuitry to perform operations comprising:

invoking the copy of the inference engine to produce the name for the content;

creating the interest packet with the name to request the content; and

transmitting the interest packet.