ACCELERATING LEARNING BY SHARING INFORMATION BETWEEN MULTIPLE LEARNING MACHINES

• US 20140222726A1
• Filed: 07/09/2013
• Published: 08/07/2014
• Est. Priority Date: 02/05/2013
• Status: Active Grant

• Current Assignee: Cisco Technology, Inc. (Cisco Systems, Inc.)
• Original Assignee: Cisco Technology, Inc. (Cisco Systems, Inc.)
• Inventors: Mermoud, Grgory, Vasseur, Jean-Philippe, Dasgupta, Sukrit
• Granted Patent: US 9,436,917 B2
• Time in Patent Office: Days
• US Class Current: 706/12
• CPC Class Codes: G06N 20/00 Machine learning
• CPC Class Codes: H04L 61/00 Network arrangements, protocols or services for addressing or naming

Abstract

In one embodiment, variables maintained by each of a plurality of Learning Machines (LMs) are determined. The LMs are hosted on a plurality of Field Area Routers (FARs) in a network, and the variables are sharable between the FARs. A plurality of correlation values defining a correlation between the variables is calculated. Then, a cluster of FARs is computed based on the plurality of correlation values, such that the clustered FARs are associated with correlated variables, and the cluster allows the clustered FARs to share their respective variables.

Claims

1. A method, comprising:

determining variables maintained by each of a plurality of Learning Machines (LMs), which are hosted on a plurality of Field Area Routers (FARs) in a network, wherein the variables are sharable between the FARs;

calculating a plurality of correlation values defining a correlation between the variables; and

computing a cluster of FARs based on the plurality of correlation values, wherein the clustered FARs are associated with correlated variables, and the cluster allows the clustered FARs to share their respective variables.

2. The method according to claim 1, further comprising:

determining whether a correlation value of the plurality of correlation values exceeds a predetermined threshold, wherein the cluster is computed only when the correlation value exceeds the predetermined threshold.

3. The method according to claim 1, wherein the determining of the variables comprises:

periodically transmitting messages to each FAR to probe the FARs for updated variables.

4. The method according to claim 1, further comprising:

determining a sharing rate defining a rate at which the variables are shared by the clustered FARs.

5. The method according to claim 4, further comprising:

determining available network resources in the network, wherein the sharing rate is determined based on the available network resources.

6. The method according to claim 4, further comprising:

receiving an indication of a user-defined sharing rate, wherein the sharing rate is determined based on the user-defined sharing rate.

7. The method according to claim 1, further comprising:

dynamically adjusting the cluster based on a network metric.

8. The method according to claim 7, further comprising:

receiving feedback from a particular LM relating to variables maintained by the particular LM, wherein the network metric is based on the received feedback.

9. The method according to claim 7, further comprising:

determining available network resources in the network, wherein the network metric is based on the available network resources.

10. An apparatus, comprising:

one or more network interfaces that communicate with a network;

a processor coupled to the one or more network interfaces and configured to execute a process; and

a memory configured to store program instructions which contain the process executable by the processor, the process comprising;

determining variables maintained by each of a plurality of Learning Machines (LMs), which are hosted on a plurality of Field Area Routers (FARs) in the network, wherein the variables are sharable between the FARs;

calculating a plurality of correlation values defining a correlation between the variables; and

computing a cluster of FARs based on the plurality of correlation values, wherein the clustered FARs are associated with correlated variables, and the cluster allows the clustered FARs to share their respective variables.

11. The apparatus according to claim 10, wherein the process further comprises:

determining whether a correlation value of the plurality of correlation values exceeds a predetermined threshold, wherein the cluster is computed only when the correlation value exceeds the predetermined threshold.

12. The apparatus according to claim 10, wherein the determining of the variables comprises:

periodically transmitting messages to each FAR to probe the FARs for updated variables.

13. The apparatus according to claim 10, wherein the process further comprises:

determining a sharing rate defining a rate at which the variables are shared by the clustered FARs.

14. The apparatus according to claim 13, wherein the process further comprises:

determining available network resources in the network, wherein the sharing rate is determined based on the available network resources.

15. The apparatus according to claim 13, wherein the process further comprises:

receiving an indication of a user-defined sharing rate, wherein the sharing rate is determined based on the user-defined sharing rate.

16. The apparatus according to claim 10, wherein the process further comprises:

dynamically adjusting the cluster based on a network metric.

17. The apparatus according to claim 16, wherein the process further comprises:

receiving feedback from an LM relating to variables maintained by the LM, wherein the network metric is based on the received feedback.

18. The apparatus according to claim 16, wherein the process further comprises:

determining available network resources in the network, wherein the network metric is based on the available network resources.

19. A tangible non-transitory computer readable medium storing program instructions that cause a computer to execute a process, the process comprising:

determining variables maintained by each of a plurality of Learning Machines (LMs), which are hosted on a plurality of Field Area Routers (FARs) in a network, wherein the variables are sharable between the FARs;

calculating a plurality of correlation values defining a correlation between the variables; and

computing a cluster of FARs based on the plurality of correlation values, wherein the clustered FARs are associated with correlated variables, and the cluster allows the clustered FARs to share their respective variables.

20. The computer readable medium according to claim 19, further comprising:

determining whether a correlation value of the plurality of correlation values exceeds a predetermined threshold, wherein the cluster is computed only when the correlation value exceeds the predetermined threshold.