Abstract: An endpoint system receives a target file for evaluation for malicious scripts. The original content of the target file is normalized and stored in a normalized buffer. Tokens in the normalized buffer are translated to symbols, which are stored in a tokenized buffer. Strings in the normalized buffer are stored in a string buffer. Tokens that are indicative of syntactical structure of the normalized content are extracted from the normalized buffer and stored in a structure buffer. The content of the tokenized buffer and counts of tokens represented as symbols in the tokenized buffer are compared against heuristic rules indicative of malicious scripts. The contents of the tokenized buffer and string buffer are compared against signatures of malicious scripts. The contents of the tokenized buffer, string buffer, and structure buffer are input to a machine learning model that has been trained to detect malicious scripts.

Abstract: A machine learning system includes multiple machine learning models. A target object, such as a file, is scanned for machine learning features. Context information of the target object, such as the type of the object and how the object was received in a computer, is employed to select a machine learning model among the multiple machine learning models. The machine learning model is also selected based on threat intelligence, such as census information of the target object. The selected machine learning model makes a prediction using machine learning features extracted from the target object. The target object is allowed or blocked depending on whether or not the prediction indicates that the target object is malicious.