Rapid, targeted network threat detection

1. A method comprising:

at a computing device having connectivity to a network, determining features of background network traffic;

extracting features from samples of a particular type of network threat;

determining a characteristic of the particular type of network threat that best differentiates the features of the particular type of network threat from the features of the background network traffic;

based on the characteristic, creating a targeted detector for the particular type of network threat that is deployable as a stand-alone detector; and

applying an action to particular incoming network traffic identified by the targeted detector as being associated with the particular type of network threat.

2. The method of claim 1 , further comprising:

determining a mean vector for the background network traffic based on the features of the background network traffic;

determining a covariance matrix for the background network traffic based on the features of the background network traffic; and

determining a mean vector for the particular type of network threat based on the features of the particular type of network threat.

3. The method of claim 2 , wherein the characteristic is based on a linear combination of a subset or all of the features of the background traffic and the features of the particular type of network threat, and determining the characteristic further comprises:

analyzing the mean vector of the particular type of network threat and the mean vector of the background traffic in view of the covariance matrix.

4. The method of claim 2 , wherein the mean vector of the particular type of network threat is the only statistical measure determined based on the features of the particular type of network threat.

5. The method of claim 1 , wherein the creating the targeted detector further comprises:

determining a detection statistic based on the characteristic, wherein the targeted detector evaluates the detection statistic of incoming network traffic based on a threshold that is set to a value that captures minimum values of the detection statistic from the samples of the particular type of network threat and identifies the particular incoming traffic as being associated with the particular type of network threat when the detection statistic of the particular incoming traffic exceeds the threshold.

6. The method of claim 1 , wherein the particular type of network threat is a particular type of malware and applying the action further comprises:

blocking traffic associated with the particular type of malware.

7. The method of claim 5 , wherein the particular type of threat is a first particular type of malware, the features of the particular type of network threat are a first set of features of the first particular type of malware, the targeted detector is a first targeted detector, the characteristic is a first characteristic, the detection statistic is a first detection statistic, the threshold is a first threshold, the action is a first action, and the method further comprises:

extracting a second set of features from samples of a second particular type of malware;

determining a second characteristic of the second particular type of malware that differentiates the second set of features from the features of the background network traffic;

determining a second detection statistic based on the second characteristic;

based on the second characteristic, creating a second targeted detector for the second particular type of malware that is deployable as a stand-alone detector and that evaluates the second detection statistic of incoming network traffic based on a second threshold, wherein the second threshold is set to a value that captures minimum values of the second detection statistic from the samples of the second particular type of malware; and

applying a second action to particular incoming network traffic identified by the second targeted detector as being associated with the second particular type of malware.

8. An apparatus comprising:

one or more network interface units that enable network connectivity; and

a processor that:

determines features of background network traffic;

extracts features from samples of a particular type of network threat;

determines a characteristic of the particular type of network threat that best differentiates the features of the particular type of network threat from the features of the background network traffic;

creates, based on the characteristic, a targeted detector for the particular type of network threat that is deployable as a stand-alone detector; and

applies an action to particular incoming network traffic identified by the targeted detector as being associated with the particular type of network threat.

9. The apparatus of claim 8 , wherein the processor further:

determines a mean vector for the background network traffic based on the features of the background network traffic;

determines a covariance matrix for the background network traffic based on the features of the background network traffic; and

determines a mean vector for the particular type of network threat based on the features of the particular type of network threat.

10. The apparatus of claim 9 , wherein the characteristic is based on a linear combination of a subset or all of the features of the background traffic and the features of the particular type of network threat, and in determining the characteristic, the processor further:

analyzes the mean vector of the particular type of network threat and the mean vector of the background traffic in view of the covariance matrix.

11. The apparatus of claim 9 , wherein the mean vector of the particular type of network threat is the only statistical measure determined based on the features of particular type of network threat.

12. The apparatus of claim 8 , wherein, in creating the targeted detector, the processor further:

determines a detection statistic based on the characteristic, wherein the targeted detector evaluates the detection statistic of incoming network traffic based on a threshold that is set to a value that captures minimum values of the detection statistic from the samples of the particular type of network threat and identifies the particular incoming traffic as being associated with the particular type of network threat when the detection statistic of the particular incoming traffic exceeds the threshold.

13. The apparatus of claim 8 , wherein the particular type of network threat is a particular type of malware and, in applying the action, the processor further:

blocks traffic associated with the particular type of network threat.

14. The apparatus of claim 12 , wherein the particular type of threat is a first particular type of malware, the features of the particular type of network threat are a first set of features of the first particular type of malware, the targeted detector is a first targeted detector, the characteristic is a first characteristic, the detection statistic is a first detection statistic, the threshold is a first threshold, the action is a first action, and the processor further:

extracts a second set of features from samples of a second particular type of malware;

determines a second characteristic of the second particular type of malware that differentiates the second set of features from the features of the background network traffic;

determines a second detection statistic based on the second characteristic;

creates, based on the second characteristic, a second targeted detector for the second particular type of malware that is deployable as a stand-alone detector and that evaluates the second detection statistic of incoming network traffic based on a second threshold, wherein the second threshold is set to a value that captures minimum values of the second detection statistic from the samples of the second particular type of malware; and

applies a second action to particular incoming network traffic identified by the second targeted detector as being associated with the second particular type of malware.

15. A non-transitory computer-readable storage media encoded with software comprising computer executable instructions and when the software is executed operable to:

determine features of background network traffic;

extract features from samples of a particular type of network threat;

determine a characteristic of the particular type of network threat that best differentiates the features of the particular type of network threat from the features of the background network traffic;

based on the characteristic, create a targeted detector for the particular type of network threat that is deployable as a stand-alone detector; and

apply an action to particular incoming network traffic identified by the targeted detector as being associated with the particular type of network threat.

16. The non-transitory computer-readable storage media of claim 15 , further comprising instructions operable to:

determine a mean vector for the background network traffic; based on the features of the background network traffic;

determine a covariance matrix for the background network traffic based on the features of the background network traffic; and

determine a mean vector for the particular type of network threat based on the features of the particular type of network threat.

17. The non-transitory computer-readable storage media of claim 16 , wherein the characteristic is based on a linear combination of a subset or all of the features of the background traffic and the features of the particular type of network threat, and the instructions operable to determine the characteristic further comprise instructions operable to:

analyze the mean vector of the particular type of network threat and the mean vector of the background traffic in view of the covariance matrix.

18. The non-transitory computer-readable storage media of claim 16 , wherein the mean vector of the particular type of network threat is the only statistical measure determined based on the features of the particular type of network threat.

19. The non-transitory computer-readable storage media of claim 15 , further comprising instructions operable to:

determine a detection statistic based on the characteristic, wherein the targeted detector evaluates the detection statistic of incoming network traffic based on a threshold that is set to a value that captures minimum values of the detection statistic from the samples of the particular type of network threat and identifies the particular incoming traffic as being associated with the particular type of network threat when the detection statistic of the particular incoming traffic exceeds the threshold; and

block traffic associated with the particular type of network threat.

20. The non-transitory computer-readable storage media of claim 15 , wherein the particular type of threat is a first particular type of malware, the features of the particular type of network threat are a first set of features of the first particular type of malware, the targeted detector is a first targeted detector, the characteristic is a first characteristic, the detection statistic is a first detection statistic, the threshold is a first threshold, the action is a first action, and the instructions are further operable to:

extract a second set of features from samples of a second particular type of malware;

determine a second characteristic of the second particular type of malware that differentiates the second set of features from the features of the background network traffic;

determine a second detection statistic based on the second characteristic;

based on the second characteristic, create a second targeted detector for the second particular type of malware that is deployable as a stand-alone detector and that evaluates the second detection statistic of incoming network traffic based on a second threshold, wherein the second threshold is set to a value that captures minimum values of the second detection statistic from the samples of the second particular type of malware; and

apply a second action to particular incoming network traffic identified by the second targeted detector as being associated with the second particular type of malware.