Abstract:Aiming at feature extraction for QAR data
a method of maximum margin discriminant analysis for one-class classification was given
and by adding orthogonality restriction on it
a maximum margin feature extraction method was put forward.For the large sample learning problem of QAR data
a modified representation for maximum margin discriminant analysis for one-class classification with orthogonality restriction was presented
by transforming the modified representation to minimum enclosing ball problem. Finally a maximum margin method of feature extraction based on minimum enclosing ball problem with orthogonality constraints was proposed
which had good performance in the real dataset from a type of aircraft
so it effectively solved the large sample feature extraction problem for QAR data.
关键词
空中交通特征提取最小闭包球快速存取纪录器
Keywords
air trafficfeature extractionminimum enclosing ballquick access recorder
references
Tsang I W;Kocsor A;Kwok J T,Efficient kernel feature extraction for massive data sets,Philadelphia,PA,USA,2006.