Network-based approach for training supervised learning classifiers
In one embodiment, a supervisory device in a network receives traffic data from a security device that uses traffic signatures to assess traffic in the network. The supervisory device receives traffic data from one or more distributed learning agents that use machine learning-based anomaly detection to assess traffic in the network. The supervisory device trains a traffic classifier using the received traffic data from the security device and from the one or more distributed learning agents. The supervisory device deploys the traffic classifier to a selected one of the one or more distributed learning agents.
1. A method comprising:
dynamically discovering, by a supervisory device in a network, a security device that uses traffic signatures to assess traffic in the network;
generating, by the supervisory device, a given rule regarding when traffic data from the security device is sent to the supervisory device;
receiving, at the supervisory device, the traffic data from the security device based on the given rule;
receiving, at the supervisory device, traffic data from one or more distributed learning agents that use machine learning-based anomaly detection to assess traffic in the network;
training, by the supervisory device, a traffic classifier using the received traffic data from the security device and from the one or more distributed learning agents; and
deploying, by the supervisory device, the traffic classifier to a selected one of the one or more distributed learning agents.
2. The method as in claim 1 , wherein the security device comprises at least one of: a firewall, an intrusion detection device, or an intrusion prevention device.
3. The method as in claim 1 , further comprising:
sending, by the supervisory device, a discovery request to the security device; and
receiving, at the supervisory device, a discovery acknowledgement from the security device in response to the discovery request, wherein the discovery acknowledgement indicates one or more capabilities of the security device.
4. The method as in claim 1 , further comprising:
sending, by the supervisory device, a traffic sample request to the security device for the traffic data received from the security device, wherein the traffic sample request indicates at least one of: a particular signature of which the requested traffic should or should not match, a traffic type, or a time at which the security device is to send the traffic data.
5. The method as in claim 1 , further comprising:
scheduling, by the supervisory device, when the one or more distributed learning agents are to send the traffic data to the supervisory device based on a topology of the network or a bandwidth utilization.
6. The method as in claim 1 , further comprising:
requesting, by the supervisory device, the traffic data received from the one or more distributed learning agents based on one or more traffic categories associated with the one or more distributed learning agents.
7. The method as in claim 1 , further comprising:
receiving, by the supervisory device, traffic data from a first agent;
generating, by the supervisory device, a statistical model using the traffic data from the first agent; and
using, by the supervisory device, the statistical model to verify whether the traffic data from the first agent is representative of traffic from a second agent.
8. The method as in claim 7 , further comprising:
requesting, by the supervisory device, the traffic data from the second agent, in response to a determination that the traffic data from the first agent is not representative of the traffic data from the second agent.
9. The method as in claim 1 , wherein the traffic classifier comprises a deep neural network.
10. An apparatus, comprising:
one or more network interfaces to communicate with a network;
a processor coupled to the network interfaces and configured to execute one or more processes; and
a memory configured to store a process executable by the processor, the process when executed operable to:
dynamically discover a security device that uses traffic signatures to assess traffic in the network;
generate a given rule regarding when traffic data from the security device is sent to the supervisory device;
receive the traffic data from the security device based on the given rule;
receive traffic data from one or more distributed learning agents that use machine learning-based anomaly detection to assess traffic in the network;
train a traffic classifier using the received traffic data from the security device and from the one or more distributed learning agents; and
deploy the traffic classifier to a selected one of the one or more distributed learning agents.
11. The apparatus as in claim 10 , wherein the security device comprises at least one of: a firewall, an intrusion detection device, or an intrusion prevention device.
12. The apparatus as in claim 10 , wherein the process when executed is further operable to:
send a discovery request to the security device; and
receive a discovery acknowledgement from the security device in response to the discovery request, wherein the discovery acknowledgement indicates one or more capabilities of the security device.
13. The apparatus as in claim 10 , wherein the process when executed is further operable to:
send a traffic sample request to the security device for the traffic data received from the security device, wherein the traffic sample request indicates at least one of: a particular signature of which the requested traffic should or should not match, a traffic type, or a time at which the security device is to send the traffic data.
14. The apparatus as in claim 10 , wherein the process when executed is further operable to:
schedule when the one or more distributed learning agents are to send the traffic data to the apparatus based on a topology of the network or a bandwidth utilization.
15. The apparatus as in claim 10 , wherein the process when executed is further operable to:
request the traffic data received from the one or more distributed learning agents based on one or more traffic categories associated with the one or more distributed learning agents.
16. The apparatus as in claim 10 , wherein the process when executed is further operable to:
receive traffic data from a first agent;
generate a statistical model using the traffic data from the first agent; and
use the statistical model to verify whether the traffic data from the first agent is representative of traffic from a second agent.
17. The apparatus as in claim 16 , wherein the process when executed is further operable to:
request the traffic data from the second agent, in response to a determination that the traffic data from the first agent is not representative of the traffic data from the second agent.
18. The apparatus as in claim 10 , wherein the traffic classifier comprises a deep neural network.
19. A tangible, non-transitory, computer-readable medium storing program instructions that cause a supervisory device in a network to execute a process comprising:
dynamically discovering, by a supervisory device in a network, a security device that uses traffic signatures to assess traffic in the network;
generating, by the supervisory device, a given rule regarding when traffic data from the security device is sent to the supervisory device;
receiving, at the supervisory device, the traffic data from the security device based on the given rule;
receiving, at the supervisory device, traffic data from one or more distributed learning agents that use machine learning-based anomaly detection to assess traffic in the network;
training, by the supervisory device, a traffic classifier using the received traffic data from the security device and from the one or more distributed learning agents; and
deploying, by the supervisory device, the traffic classifier to a selected one of the one or more distributed learning agents.
20. The tangible, non-transitory, computer-readable medium as in claim 19 , wherein the security device comprises at least one of: a firewall, an intrusion detection device, or an intrusion prevention device.