Consider 3-NN in the setting of P6 in s8_knn.pdf. Show that as n , the asymptotic error of 3-NN is upper bounded by P3nn p(1 p)(4p 4p2 + 1) (1) where p is the Bayes probability of error. Show that the bound can be weakened to P3nn 1.4p, and compare it with the case of 1-NN. Hint: Notes_knn.pdf contains the main ingredients of the analysis.