Abstract: One or more iterations are performed. Each iteration includes calculating, for each of a number of data points that each have a label probability distribution, a label quality measure based on the label probability distribution of the data point. Each iteration includes updating the label probability distribution of each of at least one of the data points using either or both of a classification technique and a constrained clustering technique based on the data points and the label quality measure of each data point.

Abstract: Command line inputs to a system by a user or automated script can comprise a number of legitimate commands but, as a series, reveal a reconnaissance attack, such as to gain knowledge of a system without a legitimate reason to do so. A trained artificial intelligence monitors the command line inputs to the system, as a series, and determines therefrom whether a match exists to a reconnaissance attack. The match may be a non-exact match, such as a match determined by a long short-term memory (LSTM) machine learning model. A reconnaissance attack response may then be initiated upon determining a match is present.

Abstract: A system includes a processor and a memory coupled with and readable by the processor and storing therein a set of instructions. When executed by the processor, the processor is caused to receive application events associated with an application and create data records based on the application events. The processor is further caused to compute an interestingness value for each of the data records based on a goal of the application, assign the computed interestingness value to each of the data records and store each of the data records with the assigned interestingness value.

Abstract: Embodiments of the present disclosure provide a system for generating risk scores in near real-time. The system includes a processor and a memory coupled with and readable by the processor and storing therein a set of instructions. When executed by the processor, the processor is caused to generate risk scores in near real-time by receiving near real-time application events associated with an application in near real-time and identifying anomalies from the near real-time application events. The processor is further caused to generate risk scores in near real-time by generating an intermediate near real-time risk score for the identified anomalies and combining the intermediate near real-time risk score with a batch risk score generated from a batch process executed prior to receiving the near real-time application events to generate a near real-time risk score.

Abstract: An apparatus may include a processor that may be caused to access a distribution of a plurality of values, each value of the plurality of values quantifying an event of an event type in a computer network. The processor may determine a mean of the plurality of values and a second highest value of the plurality of values, generate an expected maximum of the distribution based on the mean and the second highest value, and access a first value quantifying a first event of the event type in the computer network. The processor may further determine that the first event is an anomalous event based on the first value and the expected maximum.

Abstract: First-order anomaly scores are received from related anomaly detectors. Each first-order anomaly score indicates a likelihood of an anomaly at a target system. A relatedness measure of the related anomaly detectors is determined, based on the first-order anomaly scores that have been received. A higher-order anomaly score is determined based on the relatedness measure that has been determined. The higher-order anomaly score indicates a likelihood of an anomaly at the target system. An anomaly at the target system is detected based on the higher-order anomaly score.

Abstract: An apparatus may include a processor that may be caused to access a distribution of a plurality of values, each value of the plurality of values quantifying an event of an event type in a computer network. The processor may determine a mean of the plurality of values and a second highest value of the plurality of values, generate an expected maximum of the distribution based on the mean and the second highest value, and access a first value quantifying a first event of the event type in the computer network. The processor may further determine that the first event is an anomalous event based on the first value and the expected maximum.

Abstract: Humans as well as non-human actors may interact with computer devices on a computer network. As described herein, it is possible to train and apply human vs. non-human detection models to provide an indication of the probability that a human or a non-human actor was interacting with a computer device during a particular time period. The probability that a human or non-human was interacting with computers during a particular time may be used to improve various actions, including selecting one or more different threat detection models to apply during the particular time, selecting data to use with threat detection models during the time, or selecting data from the particular time to store.

Abstract: Humans as well as non-human actors may interact with computer devices on a computer network. As described herein, it is possible to train and apply human vs. non-human detection models to provide an indication of the probability that a human or a non-human actor was interacting with a computer device during a particular time period. The probability that a human or non-human was interacting with computers during a particular time may be used to improve various actions, including selecting one or more different threat detection models to apply during the particular time, selecting data to use with threat detection models during the time, or selecting data from the particular time to store.