We have a new dataset where the input data has 2 continuous variables (x, and x),...

 

Transcribed Image Text:

We have a new dataset where the input data has 2 continuous variables (x, and x), and the task is to predict a continuous value as the output y (i.e. regression). We have reason to believe the following new model for regression is a better fit to the problem domain: 4 =W + WX + WX2 i,2 A) Write down the error function for this model. You should use the sum of squared error, as in class: [5 marks] N 1 E(X)= (v -) 2N i=1 B) Derive the gradient descent update for w, using learning rate K [5 marks] C) Derive the gradient descent update for w, using learning rate K [5 marks] D) Derive the gradient descent update for w using learning rate K [5 marks] We have a new dataset where the input data has 2 continuous variables (x, and x), and the task is to predict a continuous value as the output y (i.e. regression). We have reason to believe the following new model for regression is a better fit to the problem domain: 4 =W + WX + WX2 i,2 A) Write down the error function for this model. You should use the sum of squared error, as in class: [5 marks] N 1 E(X)= (v -) 2N i=1 B) Derive the gradient descent update for w, using learning rate K [5 marks] C) Derive the gradient descent update for w, using learning rate K [5 marks] D) Derive the gradient descent update for w using learning rate K [5 marks]