Microsoft Technology Licensing, LLC

Ido Bar Av

Yonatan Most

Shai Kaplan

Anton Wolkov

A system for determining whether a velocity event is fake or real is provided. The system accesses a data store of velocity events, each of which specifies a pair of addresses that share the velocity event. For each address of the velocity events, the system sets a score for that address based on the number of addresses that share a velocity event with that address. When the score for that address satisfies an originating address criterion, the system designates that address as an originating address. The system may determine that a velocity event is real when both addresses of the velocity event are originating addresses.

Velocity event identification system

Embodiments detect unauthorized access to cloud-based resources. One technique analyzes cloud-based events to distinguish potentially malicious velocity incidents from benign velocity incidents. A velocity incident occurs when the same user causes events from two geographically remote locations in a short time. Benign velocity incidents are distinguished from malicious velocity incidents by comparing an event with past events that have the same features. Embodiments probabilistically determine if a velocity incident is malicious or benign based on a weighted multi-feature analysis. For each feature of an event, a probability is calculated based on past events that have the same feature. Then, each feature is associated with a weight based on a relative frequency of past events having that feature. A weighted average of probabilities is calculated, and the resulting probability is compared to a defined threshold to determine if the velocity incident is likely malicious or benign.

Detecting unauthorized cloud access by detecting malicious velocity incidents

An anomaly detection system is provided and includes a processor, a memory and a security application stored in the memory and including instructions. The instructions are for collecting behavior data corresponding to users of an organization accessing cloud applications. The behavior data includes parameters tracked over time for the users. The instructions are for: creating a first model based on the behavior data tracked for the users; creating a second model corresponding to a first user based on the parameters tracked for the users except the first user, where the second model excludes behavior data pertaining to the first user; scoring the second model based on the first model to generate a first score; determining whether the first user is an outlier based on the first score; and removing the behavior data corresponding to the first user from the first model if the first user is an outlier.

Growing community of inventors

Tel Aviv, Israel

Ido Bar Av