For small scale nonstationary data sets, the recently-proposed classifier time adaptive support vector machine(TASVM)
exhibits its good performance with the distinctive characteristic of simultaneously solving several subclassifiers locally
and globally. However, for large scale data sets, its high computational cost severely weakens its usefulness. In order
to overcome this shortcoming, a novel classifier named center-constrained minimal enclosing ball(CCMEB) based time
adaptive core vector machine(CCTA-CVM for brevity) for large nonstationary datasets is proposed by using core vector
machine(CVM) theory. This classifier has the merit of asymptotic linear time complexity and inherits the good performance
of TA-SVM. Experimental results show the effectiveness of the proposed classifier.