Relief算法中特征和类别的相关性是基于特征对近距离样本的区分能力。算法从训练集D中选择一个样本R，然后从和R同类的样本中寻找最近邻样本H，称为Near Hit，从和R不同类的样本中寻找最近样本M，称为Near Miss，根据以下规则更新每个特征的权重：
如果R和Near Hit在某个特征上的距离小于R和Near Miss上的距离，则说明该特征对区分同类和不同类的最近邻是有益的，则增加该特征的权重；反之，如果R和Near Hit在某个特征上的距离大于R和Near Miss上的距离，则说明该特征对区分同类和不同类的最近邻起负面作用，则降低该特征的权重。(The correlation between feature and category in Relief algorithm is based on distinguishing ability of feature to close sample. The algorithm selects a sample R from the training set D, and then searches for the nearest neighbor sample H from the samples of the same R, called Near Hit, and searches for the nearest sample M from the sample of the R dissimilar, called the Near Miss, and updates the weight of each feature according to the following rules:
If the distance between R and Near Hit on a certain feature is less than the distance between R and Near Miss, it shows that the feature is beneficial to the nearest neighbor of the same kind and dissimilar, and increases the weight of the feature; conversely, if the distance between R and Near Hit is greater than the distance on R and Near Miss, the feature is the same. The negative effect of nearest neighbor between class and different kind reduces the weight of the feature.)