Invention Grant
- Patent Title: Method and system for data classification in the presence of a temporal non-stationarity
- Patent Title (中): 存在时间非平稳性的数据分类方法和系统
-
Application No.: US12352444Application Date: 2009-01-12
-
Publication No.: US07818318B2Publication Date: 2010-10-19
- Inventor: Gideon Berger , Bhubaneswar Mishra
- Applicant: Gideon Berger , Bhubaneswar Mishra
- Applicant Address: US NY New York
- Assignee: New York University
- Current Assignee: New York University
- Current Assignee Address: US NY New York
- Agency: Dorsey & Whitney LLP
- Main IPC: G06F7/00
- IPC: G06F7/00 ; G06F17/30

Abstract:
A method and system for determining a feature of a particular pattern are provided. In particular, data records are received, and predetermined patterns that are associated with at least some of the data records are obtained. Using the system and method, particular information is extracted from at least a subset of the received data records, the particular information being indicative of the particular pattern in at least some of the data records. Then, it is determined whether the particular pattern is an unexpected pattern based on the obtained predetermined patterns. In addition, it is possible to classify and reduce data and/or parameters provided in the data records. First, the data records are received. Then, the data records which have at least one particular pattern are classified using a Multivariate Adaptive Regression Splines technique. Thereafter, the data and/or parameters of the classified data records are shrunk using a Stein's Estimator Rule technique.
Public/Granted literature
- US20090182701A1 METHOD AND SYSTEM FOR DATA CLASSIFICATION IN THE PRESENCE OF A TEMPORAL NON-STATIONARITY Public/Granted day:2009-07-16
Information query