2001c1041 Consider the following network architecture. Linear activation is used, i.e., output of the node =...

Transcribed Image Text:

2001c1041 Consider the following network architecture. Linear activation is used, i.e., output of the node = total input to the node. This network is traine with XOR dataset. Sum of squared difference is used as the loss function. 7780 dat 2022/918 w1 O x1 w2 O x2 Calculate the loss function Loss (w1,w2) for XOR and find the minimum of the loss function. Will gradient descent always be able to obtain the minimum starting at any initial solution for this problem? [5 Marks] 2001c1041 Consider the following network architecture. Linear activation is used, i.e., output of the node = total input to the node. This network is traine with XOR dataset. Sum of squared difference is used as the loss function. 7780 dat 2022/918 w1 O x1 w2 O x2 Calculate the loss function Loss (w1,w2) for XOR and find the minimum of the loss function. Will gradient descent always be able to obtain the minimum starting at any initial solution for this problem? [5 Marks]