Question 1. Classification: decide to do K-Nearest-Neighbor Given Given (x, classification. (xi, yi) for i y)...

Transcribed Image Text:

Question 1. Classification: decide to do K-Nearest-Neighbor Given Given (x, classification. (xi, yi) for i y) for i.i.d. ~ i = 1,..., n where y, is binary label, we (a) How would you choose K such that kNN a consistent classifier? Namely, what is the choice of K such that the mis-classification error can decrease to 0 as n→ ∞o? (b) Suppose (x, y) Fxy = FxFy, i.e., x and y are independent. As n→ ∞, what value shall the mis-classification error converge to? Question 1. Classification: decide to do K-Nearest-Neighbor Given Given (x, classification. (xi, yi) for i y) for i.i.d. ~ i = 1,..., n where y, is binary label, we (a) How would you choose K such that kNN a consistent classifier? Namely, what is the choice of K such that the mis-classification error can decrease to 0 as n→ ∞o? (b) Suppose (x, y) Fxy = FxFy, i.e., x and y are independent. As n→ ∞, what value shall the mis-classification error converge to?

Answer rating: 100% (QA)

Question 1 Classification Given x y B11 r b K 1 Y Axis X X then ... View the full answer