AUTOMATED FEATURE ANALYSIS, COMPARISON, AND ANOMALY DETECTION

摘要

Novel methods and systems for automated data analysis are disclosed. Data can be automatically analyzed to determine features in different applications, such as visual field analysis and comparisons. Anomalies between groups of objects may be detected through clustering of objects.

主权项

1. A computer implemented method comprising:
providing a sensing and analyzing device, the sensing and analyzing device comprising at least one sensor configured to detect at least one physical feature describing a plurality of physical objects; generating, by the sensing and analyzing device, a plurality of feature vectors representing the plurality of objects, based on the at least one physical feature, wherein the plurality of feature vectors comprises a plurality of components describing the at least one physical feature, wherein each component of the plurality of components has a numerical range, wherein the at least one physical feature is represented by at least one component within each feature vector; converting, by the sensing and analyzing device, the numerical range of each component to a range between 0 and 1, wherein the converting is carried out by formulaF1ij=OrigFij-MinFjMaxFj-MinFj where F1ij is a normalized value of ith object and jth feature component, OrigFij is an original jth feature component value of ith object, MinFj and MaxFj are a minimum value and a maximum value of jth feature component, thereby obtaining a first plurality of normalized feature vectors comprising a first plurality of normalized components and further comprising a total number of components for the at least one physical feature of each first normalized feature vector of the first plurality of normalized feature vectors; for each first normalized feature vector of the first plurality of feature vectors, dividing, by the sensing and analyzing device, the plurality of normalized components for each physical feature of each first normalized feature vector by the total number of components for each physical feature of the first normalized feature vector, thereby obtaining a second plurality of normalized feature vectors; normalizing, by the sensing and analyzing device, the second plurality of normalized feature vectors by formulaF3ij=F2ij∑j=1N(F2ij)2 where F2ij is a component of a feature vector of the second plurality of normalized feature vectors, and F3ij is a resultant component for each feature vector of a third plurality of normalized feature vectors; clustering, by the sensing and analyzing device, the third plurality of normalized feature vectors, thereby obtaining a plurality of clustered normalized feature vectors; analyzing, by the sensing and analyzing device, the plurality of physical objects, based on at least one of the first, second, or third plurality of normalized feature vectors.