Providing adversarial protection for electronic screen displays

Patent metadata

Field	Value
Publication number	US-11288408-B2
Application number	US-201916601459-A
Country	US
Kind code	B2
Filing date	Oct 14, 2019
Priority date	Oct 14, 2019
Publication date	Mar 29, 2022
Grant date	Mar 29, 2022

Abstract

Embodiments for providing adversarial protection to computing display devices by a processor. Security defenses may be provided on one or more image display devices against automated media analysis by using adversarial noise, an adversarial patch, or a combination thereof.

First claim

The invention claimed is:

1. A method, by one or more processors, for providing adversarial protection to computing display devices, comprising:
providing security defenses on one or more image display devices of a first computing device against automated media analysis by using an adversarial noise, an adversarial patch, or a combination thereof, wherein the adversarial noise, the adversarial patch, or the combination thereof is recursively applied to each of a plurality of frames generated by the one or more display devices in real-time such that each currently displayed frame rendered by the one or more display devices contains the adversarial noise, the adversarial patch, or the combination thereof at a given strength computed for the currently displayed frame; and
executing machine learning logic to perform the computing of the given strength of the adversarial noise, the adversarial patch, or the combination thereof applied to each currently displayed frame rendered by the one or more display devices, wherein the given strength is determined by implementing a feedback loop operation by the machine learning logic to analyze an output of one or more previously displayed frames captured by a second computing device.

2. The method of claim 1 , further including determining or selecting a type of the adversarial noise to implement on the one or more image display devices.

3. The method of claim 1 , further including creating or loading the adversarial patch onto the one or more image display devices.

4. The method of claim 1 , wherein analyzing the output further includes estimating an amount of which the adversarial noise, the adversarial patch, or a combination thereof affects a display quality of images output by the one or more image display devices.

5. The method of claim 1 , further including adjusting an amount of which the adversarial noise, the adversarial patch, or a combination thereof affects a display quality of images output by the one or more image display devices.

6. A system for providing adversarial protection to computing display devices, comprising:
one or more computers with executable instructions that when executed cause the system to:
provide security defenses on one or more image display devices of a first computing device against automated media analysis by using an adversarial noise, an adversarial patch, or a combination thereof, wherein the adversarial noise, the adversarial patch, or the combination thereof is recursively applied to each of a plurality of frames generated by the one or more display devices in real-time such that each currently displayed frame rendered by the one or more display devices contains the adversarial noise, the adversarial patch, or the combination thereof at a given strength computed for the currently displayed frame; and
executing machine learning logic to perform the computing of the given strength of the adversarial noise, the adversarial patch, or the combination thereof applied to each currently displayed frame rendered by the one or more display devices, wherein the given strength is determined by implementing a feedback loop operation by the machine learning logic to analyze an output of one or more previously displayed frames captured by a second computing device.

7. The system of claim 6 , wherein the executable instructions determine or select a type of the adversarial noise to implement on the one or more image display devices.

8. The system of claim 6 , wherein the executable instructions create or load the adversarial patch onto the one or more image display devices.

9. The system of claim 6 , wherein analyzing the output further includes an amount of which the adversarial noise, the adversarial patch, or a combination thereof affects a display quality of images output by the one or more image display devices.

10. The system of claim 6 , wherein the executable instructions adjust an amount of which the adversarial noise, the adversarial patch, or a combination thereof affects a display quality of images output by the one or more image display devices.

11. A computer program product, for providing adversarial protection by one or more processors, the computer program product comprising a non-transitory computer-readable storage medium having computer-readable program code portions stored therein, the computer-readable program code portions comprising:
an executable portion that provides security defenses on one or more image display devices of a first computing device against automated media analysis by using an adversarial noise, an adversarial patch, or a combination thereof, wherein the adversarial noise, the adversarial patch, or the combination thereof is recursively applied to each of a plurality of frames generated by the one or more display devices in real-time such that each currently displayed frame rendered by the one or more display devices contains the adversarial noise, the adversarial patch, or the combination thereof at a given strength computed for the currently displayed frame; and
executing machine learning logic to perform the computing of the given strength of the adversarial noise, the adversarial patch, or the combination thereof applied to each currently displayed frame rendered by the one or more display devices, wherein the given strength is determined by implementing a feedback loop operation by the machine learning logic to analyze an output of one or more previously displayed frames captured by a second computing device.

12. The computer program product of claim 11 , further including an executable portion that determines or select a type of the adversarial noise to implement on the one or more image display devices.

13. The computer program product of claim 11 , further including an executable portion that creates or loads the adversarial patch onto the one or more image display devices.

14. The computer program product of claim 11 ,
wherein analyzing the output further includes estimating an amount of which the adversarial noise, the adversarial patch, or a combination thereof affects a display quality of images output by the one or more image display devices.

15. The computer program product of claim 11 , further including an executable portion that adjusts an amount of which the adversarial noise, the adversarial patch, or a combination thereof affects a display quality of images output by the one or more image display devices.

Assignees

• IBM

Inventors

• Buesser Beat
• Nicolae Maria-Irina
• RAWAT Ambrish
• Sinn Mathieu
• Tran Ngoc Minh
• Wistuba Martin

Classifications

• G06F21/84 Primary

  output devices, e.g. displays or monitors · CPC title