$(3$ mark$)$ Implement the K$-$means$++$ algorithm introduced in the lecture and compare its performance with K$-$means on the original MNIST data set. Choose K $=10$ and compare the performance of the two algorithms in terms of clustering accuracy and adjusted random index $($ARI$).$ For the definition of these two measures $($accuracy and ARI$),$ figure it out by google or other sources.