One source of noise is error in the labels. can you propose a method to find data points that are highly likely to be mislabeled?

Rectangle methods can be used for finding data points that have chances to be mislabeled .The mechanism used in rectangle method is quite simple can easy to execute.

The easiness of the method is due to less number of parameters.The parameters help in determining the data point is inside or outside of the rectangle and the instance.

Rectangle method gives better result because of the simple structure and less variance it will produce single instances that are less effected. Thus, it can discriminate the mislabeled data better than any other wiggly structure method.