Machine Learning Model Score Obfuscation Using Step Function, Position-Dependent Noise

Patent metadata

Field	Value
Publication number	US-2020349400-A1
Application number	US-201916399677-A
Country	US
Kind code	A1
Filing date	Apr 30, 2019
Priority date	Apr 30, 2019
Publication date	Nov 5, 2020
Grant date	—

Abstract

An artefact is received. Features are extracted from this artefact which are, in turn, used to populate a vector. The vector is then input into a classification model to generate a score. The score is then modified using a step function so that the true score is not obfuscated. Thereafter, the modified score can be provided to a consuming application or process. Related apparatus, systems, techniques and articles are also described.

First claim

What is claimed is:

1 . A computer-implemented method comprising:
receiving an artefact; extracting features from the artefact and populating a vector; inputting the vector into a classification model to generate a score; modifying the score using a step function; and providing the modified score to a consuming application or process.

2 . The method of claim 1 further comprising reducing features in the vector prior to the inputting into the classification model.

3 . The method of claim 2 , wherein the features are reduced using random projection matrices.

4 . The method of claim 2 , wherein the features are reduced using principal component analysis.

5 . The method of claim 1 , wherein the classification model is a machine learning model trained using a training data set and providing a continuous scale output.

6 . The method of claim 1 , wherein the classification model characterizes the artefact as being malicious or benign to access, execute, or continue to execute.

7 . The method of claim 6 further comprising: preventing access or execution of the artefact when the classification model characterizes the artefact as being malicious.

8 . The method of claim 1 , wherein the machine learning model comprises one or more of: a logistic regression model, a neural network, a concurrent neural network, a recurrent neural network, a generative adversarial network, a support vector machine, a random forest, or a Bayesian model.

9 . The method of claim 1 , wherein the step function applies position-dependent noise to the score.

10 . A system comprising:
at least one data processor; and memory storing instructions which, when executed by the at least one data processor, result in operations comprising:
receiving an artefact;
extracting features from the artefact and populating a vector;
inputting the vector into a classification model to generate a score;
modifying the score using a step function; and
providing the modified score to a consuming application or process.

11 . The system of claim 10 , wherein the operations further comprising: reducing features in the vector prior to the inputting into the classification model.

12 . The system of claim 11 , wherein the features are reduced using random projection matrices.

13 . The system of claim 11 , wherein the features are reduced using principal component analysis.

14 . The system of claim 10 , wherein the classification model is a machine learning model trained using a training data set and providing a continuous scale output.

15 . The system of claim 10 , wherein the classification model characterizes the artefact as being malicious or benign to access, execute, or continue to execute.

16 . The system of claim 15 , wherein the operations further comprise: preventing access or execution of the artefact when the classification model characterizes the artefact as being malicious.

17 . The system of claim 10 , wherein the machine learning model comprises one or more of: a logistic regression model, a neural network, a concurrent neural network, a recurrent neural network, a generative adversarial network, a support vector machine, a random forest, or a Bayesian model.

18 . The system of claim 10 , wherein the step function applies position-dependent noise to the score.

19 . A computer-implemented method comprising:
receiving a file; extracting features from the file and populating a vector; inputting the vector into a classification model to generate a score, the classification model being a machine learning model trained to characterize a likelihood of the file as being malicious; obfuscating the score using a step function; and providing the modified score to a consuming application or process.

20 . The method of claim 19 further comprising:
preventing access or execution of the artefact when the classification model characterizes the artefact as being malicious.

Assignees

• Cylance Inc

Inventors

• Buckingham Hailey
• Beveridge David N

Classifications

• G06F21/562 Primary

  Static detection · CPC title

• G06F18/2411

  based on the proximity to a decision surface, e.g. support vector machines · CPC title

• G06F18/2135

  based on approximation criteria, e.g. principal component analysis · CPC title

• G06F18/24

  Classification techniques · CPC title

• G06N3/0475

  Generative networks · CPC title