US 12,468,953 B2
Self-learning and adapting cyber threat defense
Tim Uwe Scheideler, Schoenenberg (CH); Arjun Udupi Raghavendra, Zurich (CH); Matthias Seul, Pleasant Hill, CA (US); and Andrea Giovannini, Zurich (CH)
Assigned to Kyndryl Inc., New York, NY (US)
Filed by Kyndryl, Inc., New York, NY (US)
Filed on May 30, 2024, as Appl. No. 18/678,415.
Application 18/678,415 is a continuation of application No. 17/249,133, filed on Feb. 22, 2021, granted, now 12,039,455.
Prior Publication US 2024/0320499 A1, Sep. 26, 2024
This patent is subject to a terminal disclaimer.
Int. Cl. G06N 3/088 (2023.01); G06F 21/56 (2013.01); G06N 3/045 (2023.01)
CPC G06N 3/088 (2013.01) [G06F 21/56 (2013.01); G06N 3/045 (2023.01); G06F 2221/034 (2013.01)] 20 Claims
OG exemplary drawing
 
1. A computer-implemented method for building a multi-category machine-learning system for detecting malware, the method comprising:
training a first machine-learning system, comprising a neural network that includes a generator portion and a discriminator portion, by using known malware patterns and related malware categories as training data until the discriminator portion is unable to differentiate between known malware patterns and synthetic code patterns;
training a second machine-learning system by using benevolent code patterns and additional synthetic code patterns, generated using the trained first machine-learning system, as training data until the second machine-learning system is enabled to predict malicious code patterns of the additional synthetic code patterns and related categories of the additional synthetic code patterns;
determining a statistical distribution of predicted malicious code patterns and related categories; and
determining a quality value of the training of the second machine-learning system, wherein the quality value denotes an indicator of a prediction accuracy of the second machine-learning system for predicting malware.