I was studying about clustering algorithms like K-means, C-means etc. While studying the same, I had the following doubt :
We are given some points in 2-dimensional space. It is guaranteed that these points can be divided into 2 clusters, meaning they are linearly separable. It is also given what point belongs to which cluster too. But there is some noise in the data, meaning that presence of such points doesn't allow us to draw a line which separates the given 2 types of points. We need to find minimum number of points which are noisy points in the data. Is there an efficient way to detect the same?
In simple words, We are given N points and the clusters to which they belong too. But some points are given as noise in the data which need to be removed so that the classified points can become linearly separable. We need to find the minimum number of noisy points.
Thanks in advance.