Data points一多,KNN会巨慢。要是我就这么说,first ask if the data has high dimensionality. If they say no, well, we can use KD-tree to speed up K-NN. If they say yes, I'd say you cannot use the original K-NN, you will have to cut corners. for example, instead of using all data points, only pick the most important ones (close to decision boundary). see https://link.springer. com/article/10.1023%2FA%3A1007626913721?LI=true
【在 z*******h 的大作中提到】 : Data points一多,KNN会巨慢。要是我就这么说,first ask if the data has high : dimensionality. If they say no, well, we can use KD-tree to speed up K-NN. : If they say yes, I'd say you cannot use the original K-NN, you will have to : cut corners. for example, instead of using all data points, only pick the : most important ones (close to decision boundary). see https://link.springer. : com/article/10.1023%2FA%3A1007626913721?LI=true