How to Use Soft Crashering in Machine Learning?

Soft crashing is actually more accurate than hard crashing. In low dimensional space, if the space is cannd for a large number of dimensional space, then those data points are very close. This means te it is not so easy to use soft crashering technique to candle for clastein those data points. And when you generate some results with your methods, how do you go about evaluating them?