Identifying malware-suspect end points through entropy changes in consolidated logs
Example methods disclosed herein to determine whether a first monitored device is compromised include determining a first entropy value for the first monitored device based on a first number of unique event identifiers included in log entries obtained for the first monitored device, the log entries associated with a first time window. Disclosed example methods also include determining a second entropy value for the first monitored device based on numbers of unique event identifiers included in corresponding groups of log entries obtained for respective ones of a plurality of monitored devices including the first monitored device, the groups of log entries associated with the first time window. Disclosed example methods further include determining whether the first monitored device is compromised based on the first entropy value and the second entropy value, and performing an action in response to a determination that the first monitored device is compromised.
1. At least one machine readable medium implemented by at least one of a solid-state device or a storage disk, the at least one machine readable medium comprising instructions, to cause one or more processors to at least:
process monitored log entries associated with a plurality of monitored devices to determine measurements associated with the monitored devices, the measurements including ones of a first type of entropy value and ones of a second type of entropy value associated respectively with the monitored devices, the ones of the first type of entropy value based on respective numbers of unique event identifiers included in respective groups of the monitored log entries associated respectively with the monitored devices, a first one of the second type of entropy value based on (i) a number of unique event identifiers included in a first one of the groups of the monitored log entries associated with a first one of the monitored devices and (ii) a total number of log entries included in the first one of the groups of the monitored log entries associated with the first one of the monitored devices;
determine, based on the measurements and a machine learning algorithm, whether the first one of the monitored devices is compromised; and
quarantine the first one of the monitored devices in response to a determination that the first one of the monitored devices is compromised.
2. The at least one machine readable medium of claim 1 , wherein the instructions are to cause the one or more processors to execute the machine learning algorithm to classify the first one of the monitored devices as compromised or uncompromised based on the measurements.
3. The at least one machine readable medium of claim 1 , wherein the instructions are to cause the one or more processors to:
execute the machine learning algorithm to determine a threshold based on the monitored log entries; and
compare a first one of the first type of entropy value to the threshold to determine whether the first one of the monitored devices is compromised, the first one of the first type of entropy value associated with the first one of the monitored devices.
4. The at least one machine readable medium of claim 1 , wherein the measurements include first confidence values based on the ones of the first type of entropy value and second confidence values based on the ones of the second type of entropy value.
5. The at least one machine readable medium of claim 1 , wherein the groups of the monitored log entries are associated with a time window.
6. The at least one machine readable medium of claim 1 , wherein the instructions are to cause the one or more processors to train the machine learning algorithm based on historical log data.
7. An apparatus comprising:
at least one memory;
computer readable instructions; and
processor circuitry to execute the computer readable instructions to at least:
process monitored log entries associated with a plurality of monitored devices to determine measurements associated with the monitored devices, the measurements including ones of a first type of entropy value and ones of a second type of entropy value associated respectively with the monitored devices, the ones of the first type of entropy value based on respective numbers of unique event identifiers included in respective groups of the monitored log entries associated respectively with the monitored devices, a first one of the second type of entropy value based on (i) a number of unique event identifiers included in a first one of the groups of the monitored log entries associated with a first one of the monitored devices and (ii) a total number of log entries included in the first one of the groups of the monitored log entries associated with the first one of the monitored devices;
determine, based on the measurements and a machine learning algorithm, whether the first one of the monitored devices is compromised; and
quarantine the first one of the monitored devices in response to a determination that the first one of the monitored devices is compromised.
8. The apparatus of claim 7 , wherein the processor circuitry is to execute the machine learning algorithm to classify the first one of the monitored devices as compromised or uncompromised based on the measurements.
9. The apparatus of claim 7 , wherein the processor circuitry is to:
execute the machine learning algorithm to determine a threshold based on the monitored log entries; and
compare a first one of the first type of entropy value to the threshold to determine whether the first one of the monitored devices is compromised, the first one of the first type of entropy value associated with the first one of the monitored devices.
10. The apparatus of claim 7 , wherein the measurements include first confidence values based on the ones of the first type of entropy value and second confidence values based on the ones of the second type of entropy value.
11. The apparatus of claim 7 , wherein the groups of the monitored log entries are associated with a time window.
12. The apparatus of claim 7 , the processor circuitry is to train the machine learning algorithm based on historical log data.
13. A method comprising:
processing, by executing an instructions with at least one processor, monitored log entries associated with a plurality of monitored devices to determine measurements associated with the monitored devices, the measurements including ones of a first type of entropy value and ones of a second type of entropy value associated respectively with the monitored devices, the ones of the first type of entropy value based on respective numbers of unique event identifiers included in respective groups of the monitored log entries associated respectively with the monitored devices, a first one of the second type of entropy value based on (i) a number of unique event identifiers included in a first one of the groups of the monitored log entries associated with a first one of the monitored devices and (ii) a total number of log entries included in the first one of the groups of the monitored log entries associated with the first one of the monitored devices;
determining, based on the measurements and a machine learning algorithm, whether the first one of the monitored devices is compromised; and
quarantining the first one of the monitored devices in response to a determination that the first one of the monitored devices is compromised.
14. The method of claim 13 , further including executing the machine learning algorithm to classify the first one of the monitored devices as compromised or uncompromised based on the measurements.
15. The method of claim 13 , further including:
executing the machine learning algorithm to determine a threshold based on the monitored log entries; and
comparing a first one of the first type of entropy value to the threshold to determine whether the first one of the monitored devices is compromised, the first one of the first type of entropy value associated with the first one of the monitored devices.
16. The method of claim 13 , wherein the measurements include first confidence values based on the ones of the first type of entropy value and second confidence values based on the ones of the second type of entropy value.
17. The method of claim 13 , wherein the groups of the monitored log entries are associated with a time window.