一个高效的k-d tree实现，可以广泛的应用与图形图像处理。-K nearest neighbor search based on KD-tree

自适应空间球的k邻域快速搜索算法，利用空间微分块与动态球策略的k近邻搜索算法研究-Origianl-Adaptive spatial neighborhood k ball fast search algorithm, differential block the use of space and dynamic ball strategy k nearest neighbor search algorithm-Origianl

该代码包实现了直推学习算法(transductive learning),该算法基于kNN的一个扩展。优点是不需要任何启发式诱导，可以防止不收敛或者局部收敛-We present a new method for transductive learning, which can be seen as a transductive version of the k nearest-neighbor classifier.