公开(公告)号：US20240289255A1

公开(公告)日：2024-08-29

申请号：US18573901

申请日：2021-06-25

申请人： Red Bend Ltd.

发明人： Shachar MENDELOWITZ ,  Jonathan KATZ ,  Ori GOLDBERG ,  Yosef GOLAN

IPC分类号： G06F11/36 ,  G06N3/08

CPC分类号： G06F11/3636 ,  G06N3/08

摘要： Disclosed herein are methods and systems for training and using a neural network to evaluate vulnerability of software packages, comprising using a plurality of training samples each associating one of a plurality of software packages with one of a plurality of vulnerabilities identified by one of a plurality of validators to training the neural network to compute a probability of presence of one or more of the plurality of vulnerabilities in each of the plurality of software packages and outputting the trained neural network. The validators may include expert knowledge, heuristics, rule-based models as well as machine learning and deep learning models. The trained neural network may be then applied to compute a probability of presence of one or more of the vulnerabilities in one or more previously unseen software packages based on a feed of vulnerabilities identified in the previously unseen software package(s) by the plurality of validators.