71 results about "Computational statistics" patented technology

In embodiments of the present invention improved capabilities are described for predicting the reputation of a communication identifier, such as a web address, a domain name, an IP address, host name, email address, IM address, telephone number, VoIP telephony address, and the like. In embodiments, the present invention may receive a communication from a first communication identifier, parse the first communication identifier into its components, and assign the components to a hierarchical tree structure, where the hierarchical tree structure maintains the hierarchical relationship between the components of the communication identifier. The present invention may monitor and keep count of a number of communications from the first communication identifier, wherein the number of communications may be kept for both malicious and / or unwanted communications and non-malicious and / or unwanted communications. Attributes may then be provided to the number of communications for each appropriate component of the hierarchical tree, and a statistical measure may be calculated as related to the number of communications for each component of the hierarchical tree. The present invention may then receive a communication from a second communication identifier, where the second communication identifier may be previously unknown and have a common component with the hierarchical tree. The statistical measure of the common component may then be assigned to the second communication identifier, and utilizing the statistical measure assigned to the second communication identifier, may provide a prediction of reputation of the second communication identifier.

A method for estimating the performance of a statistical classifier. The method includes inputting a first set of business data in a first format from a real business process and storing the first set of business data in the first format into memory. The method applying a statistical classifier to the first set of business data and recording its classification decisions and obtaining a labeling that contains the correct decision for each data item. The method includes computing a weight for each data item that reflects its true frequency and computing a performance measure of the statistical classifier based on the weights that reflect true frequency. The method also displays the performance measure to a user.

In one embodiment, traffic data that originates from sensors, cameras, or observations is analyzed. The traffic data is associated with multiple repeating time epochs or intervals. The traffic data is divided into clusters using a clustering technique. The clustering technique may include clusters of variable sizes. Each of the clusters is analyzed to calculate statistical parameters including but not limited to an average value for one or more clusters and a standard deviation value for one or more clusters. In response to a request for traffic data, simulated traffic data may be generated by providing the average value and the standard deviation value for one or more of the clusters.

Embodiments of the present invention provide devices and techniques for detecting anomalies in monitored environments. In one embodiment of the invention, data may be acquired by monitoring an environmental variable. Based on the acquired data, statistics may be calculated which model the behavior of the environmental variable. Based on the calculated statistics, a long-term behavior and a short-term behavior of the statistics may also be calculated. A difference between the long-term statistics and the short-term statistics may then be calculated. If the difference between the long-term statistical behavior and short-term statistical behavior exceeds a dynamic or predefined threshold, an action may be taken.