Abstract: In one aspect, a computer-implemented method is disclosed. The computer-implemented method may include determining a sketch matrix that approximates a matrix representative of a reference dataset. The reference dataset may include at least one computer program having a predetermined classification. A reduced dimension representation of the reference dataset may be generated based at least on the sketch matrix. The reduced dimension representation may have a fewer quantity of features than the reference dataset. A target computer program may be classified based on the reduced dimension representation. The target computer program may be classified to determine whether the target computer program is malicious. Related systems and articles of manufacture, including computer program products, are also disclosed.

Abstract: A nested file having a primary file and at least one secondary file embedded therein is parsed using at least one parser of a cell. The cell assigns a maliciousness score to each of the parsed primary file and each of the parsed at least one secondary file. Thereafter, the cell generates an overall maliciousness score for the nested file that indicates a level of confidence that the nested file contains malicious content. The overall maliciousness score is provided to a data consumer indicating whether to proceed with consuming the data contained within the nested file.

Abstract: An identity of a user on a first computing node of a plurality of nodes within a computing environment is authenticated. A first authentication score for the user is calculated at the first computing node using at least one machine learning model. The first authentication score characterize interactions of the user with the first computing node. Subsequent to such authentication, traversal of the user from the first computing node to other computing nodes among the plurality of computing nodes are monitored. An authentication score characterizing interactions of the user with the corresponding computing node are calculated at each of the nodes using respective machine learning models executing on such nodes The respective machine learning models use, as an attribute, an authentication score calculated at a previously traversed computing node. Thereafter, an action is initiated at one of the computing nodes based on the calculated authentication scores.

Abstract: A machine learning model is applied to at least determine whether a computer program includes vulnerable code. The machine learning model is trained to determine whether the computer program includes vulnerable code based at least on a presence and/or absence of a first trait. An indication can be provided, via a user interface, an indication that the computer program includes vulnerable code, when the computer program is determined to include vulnerable code. Related methods and articles of manufacture, including computer program products, are also provided.

Abstract: In one respect, there is provided a system for classifying an instruction sequence with a machine learning model. The system may include at least one processor and at least one memory. The memory may include program code that provides operations when executed by the at least one processor. The operations may include: processing an instruction sequence with a trained machine learning model configured to detect one or more interdependencies amongst a plurality of tokens in the instruction sequence and determine a classification for the instruction sequence based on the one or more interdependencies amongst the plurality of tokens; and providing, as an output, the classification of the instruction sequence. Related methods and articles of manufacture, including computer program products, are also provided.

Abstract: An artefact is received. Features from such artefact are extracted and then populated in a vector. Subsequently, one of a plurality of available dimension reduction techniques are selected. Using the selected dimension reduction technique, the features in the vector are reduced. The vector is then input into a classification model and the score can be provided to a consuming application or process. Related apparatus, systems, techniques and articles are also described.

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.

Abstract: An endpoint computer system monitors data relating to a plurality of events occurring within an operating environment of the endpoint computer system. The monitoring can include receiving and/or inferring the data using one or more sensors executing on the endpoint computer system. The endpoint computer system can store artifacts used in connection with the plurality of events in a vault maintained on such endpoint computer system. The endpoint computer system, in response to a trigger, identifies and retrieves metadata characterizing artifacts associated with the trigger from the vault. Such identified and retrieved metadata is then provided by the endpoint computer system to a remote server.

Abstract: In one respect, there is provided a system for training a neural network adapted for classifying one or more instruction sequences. The system may include at least one processor and at least one memory. The memory may include program code which when executed by the at least one processor provides operations including: training, based at least on training data, a machine learning model to detect one or more predetermined interdependencies amongst a plurality of tokens in the training data; and providing the trained machine learning model to enable classification of one or more instruction sequences. Related methods and articles of manufacture, including computer program products, are also provided.

Abstract: Hash-based application programming interface (API) importing can be prevented by allocating a name page and a guard page in memory. The name page and the guard page being associated with (i) an address of names array, (ii) an address of name ordinal array, and (iii) an address of functions array that are all generated by an operating system upon initiation of an application. The name page can then be filled with valid non-zero characters. Thereafter, protections on the guard page can be changed to no access. An entry is inserted into the address of names array pointing to a relative virtual address corresponding to anywhere within the name page. Access to the guard page causes the requesting application to terminate. Related apparatus, systems, techniques and articles are also described.

Abstract: In one respect, there is provided a system for training a neural network adapted for classifying one or more scripts. The system may include at least one processor and at least one memory. The memory may include program code that provides operations when executed by the at least one memory. The operations may include: extracting, from an icon associated with a file, one or more features; assigning, based at least on the one or more features, the icon to one of a plurality of clusters; and generating, based at least on the cluster to which the icon is assigned, a classification for the file associated with the icon. Related methods and articles of manufacture, including computer program products, are also provided.

Abstract: Data is received or accessed that includes a structured file encapsulating data required by an execution environment to manage executable code wrapped within the structured file. Thereafter, code and data regions are iteratively identified in the structured file. Such identification is analyzed so that at least one feature can be extracted from the structured file. Related apparatus, systems, techniques and articles are also described.

Abstract: Executable memory space is protected by receiving, from a process, a request to configure a portion of memory with a memory protection attribute that allows the process to perform at least one memory operation on the portion of the memory. Thereafter, the request is responded to with a grant, configuring the portion of memory with a different memory protection attribute than the requested memory protection attribute. The different memory protection attribute restricting the at least one memory operation from being performed by the process on the portion of the memory. In addition, it is detected when the process attempts, in accordance with the grant, the at least one memory operation at the configured portion of memory. Related systems and articles of manufacture, including computer program products, are also disclosed.

Abstract: Described are techniques to enable computers to efficiently determine if they should run a program based on an immediate (i.e., real-time, etc.) analysis of the program. Such an approach leverages highly trained ensemble machine learning algorithms to create a real-time discernment on a combination of static and dynamic features collected from the program, the computer's current environment, and external factors. Related apparatus, systems, techniques and articles are also described.

Abstract: Each of a plurality of endpoint computer systems monitors data relating to a plurality of events occurring within an operating environment of the corresponding endpoint computer system. The monitoring can include receiving and/or inferring the data using one or more sensors executing on the endpoint computer systems Thereafter, for each endpoint computer system, artifacts used in connection with the events are stored in a vault maintained on such endpoint computer system. A query is later received by at least a subset of the plurality of endpoint computer systems from a server. Such endpoint computer systems, in response, identify and retrieve artifacts within the corresponding vaults response to the query. Results responsive to the query including or characterizing the identified artifacts is then provided by the endpoint computer systems receiving the query to the server.

Abstract: Centroids are used for improving machine learning classification and information retrieval. A plurality of files are classified as malicious or not malicious based on a function dividing a coordinate space into at least a first portion and a second portion such that the first portion includes a first subset of the plurality of files classified as malicious. One or more first centroids are defined in the first portion that classify files from the first subset as not malicious. A file is determined to be malicious based on whether the file is located within the one or more first centroids.

Abstract: Under one aspect, a method is provided for protecting a device from a malicious file. The method can be implemented by one or more data processors forming part of at least one computing device and can include extracting from the file, by at least one data processor, sequential data comprising discrete tokens. The method also can include generating, by at least one data processor, n-grams of the discrete tokens. The method also can include generating, by at least one data processor, a vector of weights based on respective frequencies of the n-grams. The method also can include determining, by at least one data processor and based on a statistical analysis of the vector of weights, that the file is likely to be malicious. The method also can include initiating, by at least one data processor and responsive to determining that the file is likely to be malicious, a corrective action.

Abstract: Methods are provided herein for communications bus signal fingerprinting. A security module monitors a plurality of voltage lines of at least one electronic control unit (ECU) electrically coupled to a communications bus. A voltage differential across at least two of the plurality of voltage lines of the at least one ECU is measured. The voltage differential is compared to a plurality of predetermined signal fingerprints associated with the at least one ECU. A variance in the compared voltage differential is identified relative to one or more of the plurality of predetermined signal fingerprints. Data characterizing the identified variance is provided.

Abstract: A plurality of events associated with each of a plurality of computing nodes that form part of a network topology are monitored. The network topology includes antivirus tools to detect malicious software prior to it accessing one of the computing nodes. Thereafter, it is determined that, using at least one machine learning model, at least one of the events is indicative of malicious activity that has circumvented or bypassed the antivirus tools. Data is then provided that characterizes the determination. Related apparatus, systems, techniques and articles are also described.

Abstract: Using a recurrent neural network (RNN) that has been trained to a satisfactory level of performance, highly discriminative features can be extracted by running a sample through the RNN, and then extracting a final hidden state hh where i is the number of instructions of the sample. This resulting feature vector may then be concatenated with the other hand-engineered features, and a larger classifier may then be trained on hand-engineered as well as automatically determined features. Related apparatus, systems, techniques and articles are also described.