Abstract: The K-Nearest Neighbors (KNN) algorithm is a classical supervised learning method widely used in classification and regression ... In this study, an optimization strategy based on balanced ...

Abstract: K-nearest neighbor (K-NN) algorithm is a classification method based on statistical theory. In this algorithm the Euclidean distance is usually chosen as the similarity measure, which ...

With the help of real-valued EA algorithms, direct optimization of the spatial filter with respect to the classification accuracy can be performed where the convexity may not be preserved in the cost ...

Methods: The INGHS algorithm is applied to ... frequency-temporal parameters in the optimization process so that the CSP features obtained by the optimal frequency-time parameters can enhance the ...