Given input data X = {x1, x2,...,xN} for x; ERP and labels y = {1, 2,...,...

   

Transcribed Image Text:

Given input data X = {x1, x2,...,xN} for x; ERP and labels y = {1, 2,..., YN} for y; R, L1- regression finds the linear model = wx+6 by minimizing the L1 norm between prediction and ground- truth label y where is the absolute value. Q1 main - (w,b) Let N = 1, x = [2] 1 = 0, w answer to one decimal place (e.g. 0.2): aw and b = 0. Compute the following gradients, and round your ab Q2 Derive the the gradient of the loss with respect to b, assuming y; -wx-b0 for all 1 Q3 Derive the the gradient of the loss l with respect to w, assuming yi - wTx-b0 for all 1 Given input data X = {x1,x2,...,xN} for x, ERD and labels y = {y1, 92, 9N} for y, R, L2- regression finds the linear model = w x + b by minimizing the L2 norm between prediction and ground- truth label y Q4 min w,b (wx +b-y). (w,b) Let N = 1,x = [2]. Y = 0, w = [A] and b = 0. Compute the following gradients, and round your answer to one decimal place (e.g. 0.2): = Given input data X = {x1, x2,...,xN} for x; ERP and labels y = {1, 2,..., YN} for y; R, L1- regression finds the linear model = wx+6 by minimizing the L1 norm between prediction and ground- truth label y where is the absolute value. Q1 main - (w,b) Let N = 1, x = [2] 1 = 0, w answer to one decimal place (e.g. 0.2): aw and b = 0. Compute the following gradients, and round your ab Q2 Derive the the gradient of the loss with respect to b, assuming y; -wx-b0 for all 1 Q3 Derive the the gradient of the loss l with respect to w, assuming yi - wTx-b0 for all 1 Given input data X = {x1,x2,...,xN} for x, ERD and labels y = {y1, 92, 9N} for y, R, L2- regression finds the linear model = w x + b by minimizing the L2 norm between prediction and ground- truth label y Q4 min w,b (wx +b-y). (w,b) Let N = 1,x = [2]. Y = 0, w = [A] and b = 0. Compute the following gradients, and round your answer to one decimal place (e.g. 0.2): =