2. We want to find the optimal parameter w. € Rfor linear regression with weighted mean squared error and the data below
Posted: Thu Apr 28, 2022 7:41 am
2. We want to find the optimal parameter w. € Rfor linear regression with weighted mean squared error and the data below: x-(!?).7-6 = 3 4 5 3 2 6 2 y = 17, weight k = 4), w, 2 -8--0) 3 That is, we want to find w., such that W. = argminy 'N į )? ki(w'x; -y:) i=1 where k; is the weight for i-th data sample. (a) Derive a gradient algorithm (steepest descent) with the learning rate 0.005, and continue the steps until t = 5 by using your algorithm and initial guess w, above. That is, you have to compute and show w; for 1st s 5. (b) Derive the closed form solution and find w. with it.