MALWARE DETECTION ON ENCRYPTED DATA
Methods, systems, and devices for data management are described. A data management system (DMS) may support the detection of malware on encrypted data associated with a computing system that is backed up by the DMS. For example, the DMS may transmit first signaling that instructs the computing system to generate and encrypt one or more machine learning (ML) model features associated with a computing object of the computing system. In response, the DMS may receive second signaling from the computing system that includes the one or more encrypted ML model features. The DMS may use an ML model and the one or more encrypted ML model features as inputs to the ML model to generate and transmit, to the computing system, an encrypted indication of whether malware is present on the computing object.
1 . A method for data management, comprising:
transmitting, by a data management system to a computing system backed up by the data management system, first signaling that instructs the computing system to generate and encrypt one or more machine learning model features associated with a computing object of the computing system;
receiving, at the data management system from the computing system, second signaling comprising the one or more encrypted machine learning model features;
generating, using a machine learning model and the one or more encrypted machine learning model features as inputs to the machine learning model, an encrypted indication of whether malware is present on the computing object; and
transmitting, from the data management system to the computing system, the encrypted indication.
2 . The method of claim 1 , further comprising:
receiving, at the data management system, unencrypted machine learning model features; and
using the unencrypted machine learning model features to train the machine learning model to detect the presence of malware on the computing object, wherein generating the encrypted indication is based at least in part on training the machine learning model using the unencrypted machine learning model features.
3 . The method of claim 2 , wherein:
the one or more machine learning model features are encrypted according to a homomorphic encryption scheme, and
the machine learning model trained using unencrypted machine learning model features is used to detect whether malware is present on the computing object using the one or more encrypted machine learning model features as inputs based at least in part on the one or more machine learning model features being encrypted according to the homomorphic encryption scheme.
4 . The method of claim 1 , further comprising:
refraining, at the data management system, from accessing data of the computing object in association with determining whether malware is present on the computing object based at least in part on generating the encrypted indication using the one or more encrypted machine learning model features.
5 . The method of claim 1 , wherein presence or absence of malware on the computing object is unknown to the data management system based at least in part on using the one or more encrypted machine learning model features to generate the encrypted indication.
6 . The method of claim 1 , wherein the one or more machine learning model features comprise an identifier associated with an application programming interface called by the computing object, a file size of the computing object, one or more changes to the file size of the computing object, a file header of the computing object, a time of creation of the computing object, an entropy associated with the computing object, packer information associated with the computing object, one or more parameters associated with an operating system of the computing system, or any combination thereof.
7 . The method of claim 1 , wherein the indication of whether malware is present on the computing object is encrypted based at least in part on using the one or more encrypted machine learning model features as the inputs to the machine learning model.
8 . The method of claim 1 , wherein:
the one or more machine learning model features are encrypted based at least in part on a public key associated with the computing system, and
the encrypted indication is decrypted based at least in part on a private key associated with the computing system.
9 . An apparatus for data management, comprising:
at least one processor;
memory coupled with the at least one processor; and
instructions stored in the memory and executable by the at least one processor to cause the apparatus to:
transmit, by a data management system to a computing system backed up by the data management system, first signaling that instructs the computing system to generate and encrypt one or more machine learning model features associated with a computing object of the computing system;
receive, at the data management system from the computing system, second signaling comprising the one or more encrypted machine learning model features;
generate, using a machine learning model and the one or more encrypted machine learning model features as inputs to the machine learning model, an encrypted indication of whether malware is present on the computing object; and
transmit, from the data management system to the computing system, the encrypted indication.
10 . The apparatus of claim 9 , wherein the instructions are further executable by the at least one processor to cause the apparatus to:
receive, at the data management system, unencrypted machine learning model features; and
used the unencrypted machine learning model features to train the machine learning model to detect the presence of malware on the computing object, wherein generating the encrypted indication is based at least in part on training the machine learning model using the unencrypted machine learning model features.
11 . The apparatus of claim 10 , wherein:
the one or more machine learning model features are encrypted according to a homomorphic encryption scheme, and
the machine learning model trained using unencrypted machine learning model features is used to detect whether malware is present on the computing object using the one or more encrypted machine learning model features as inputs based at least in part on the one or more machine learning model features being encrypted according to the homomorphic encryption scheme.
12 . The apparatus of claim 9 , wherein presence or absence of malware on the computing object is unknown to the data management system based at least in part on using the one or more encrypted machine learning model features to generate the encrypted indication.
13 . The apparatus of claim 9 , wherein the one or more machine learning model features comprise an identifier associated with an application programming interface called by the computing object, a file size of the computing object, one or more changes to the file size of the computing object, a file header of the computing object, a time of creation of the computing object, an entropy associated with the computing object, packer information associated with the computing object, one or more parameters associated with an operating system of the computing system, or any combination thereof.
14 . The apparatus of claim 9 , wherein the indication of whether malware is present on the computing object is encrypted based at least in part on using the one or more encrypted machine learning model features as the inputs to the machine learning model.
15 . The apparatus of claim 9 , wherein:
the one or more machine learning model features are encrypted based at least in part on a public key associated with the computing system, and
the encrypted indication is decrypted based at least in part on a private key associated with the computing system.
16 . A non-transitory computer-readable medium storing code for data management, the code comprising instructions executable by at least one processor to:
transmit, by a data management system to a computing system backed up by the data management system, first signaling that instructs the computing system to generate and encrypt one or more machine learning model features associated with a computing object of the computing system;
receive, at the data management system from the computing system, second signaling comprising the one or more encrypted machine learning model features;
generate, using a machine learning model and the one or more encrypted machine learning model features as inputs to the machine learning model, an encrypted indication of whether malware is present on the computing object; and
transmit, from the data management system to the computing system, the encrypted indication.
17 . The non-transitory computer-readable medium of claim 16 , wherein the instructions are further executable by the at least one processor to:
receive, at the data management system, unencrypted machine learning model features; and
used the unencrypted machine learning model features to train the machine learning model to detect the presence of malware on the computing object, wherein generating the encrypted indication is based at least in part on training the machine learning model using the unencrypted machine learning model features.
18 . The non-transitory computer-readable medium of claim 17 , wherein:
the one or more machine learning model features are encrypted according to a homomorphic encryption scheme, and
the machine learning model trained using unencrypted machine learning model features is used to detect whether malware is present on the computing object using the one or more encrypted machine learning model features as inputs based at least in part on the one or more machine learning model features being encrypted according to the homomorphic encryption scheme.
19 . The non-transitory computer-readable medium of claim 16 , wherein presence or absence of malware on the computing object is unknown to the data management system based at least in part on using the one or more encrypted machine learning model features to generate the encrypted indication.
20 . The non-transitory computer-readable medium of claim 16 , wherein the one or more machine learning model features comprise an identifier associated with an application programming interface called by the computing object, a file size of the computing object, one or more changes to the file size of the computing object, a file header of the computing object, a time of creation of the computing object, an entropy associated with the computing object, packer information associated with the computing object, one or more parameters associated with an operating system of the computing system, or any combination thereof.