# How does regularization work?

Could any explain how regularization works with the noise point. Like in this image, the $$2^{nd}$$ graph is where the decision surface is good, where the noise points are close to the actual points.