Runtime Optimization of Widrow-Haff Classification Algorithm Using Proper Learning Samples

This study works on the runtime optimization of Widrow-Hoff classification algorithm. The use of proper learning samples has a significant effect on the runtime and accuracy of supervised classification algorithms, in special Widrow-Hoff classification algorithm. In this study with synthesizing Multi Class Instance Selection (MCIS) algorithm and Widrow-Hoff classification algorithm, the runtime of algorithm has significantly reduced. Results of this, of vantage sample of accuracy and time, have been assessed, and simulations are indicating MCIS with the aid of proper measures is able to select the data having most effectiveness on classification. In the case, if Widrow-Hoff classifier has less and important samples (achieved by MCIS), it would be able to save significant amount of time in classification process.