answer all subparts and draw the draw decision boundaries as well by HAND. 

Q1 subparts -

 

 

 

Q4 --PLEASE SOLVE ALL PARTS AND DO EVERYTHING TRYING TO DO ON PAPER : 

  
 
 
 

Transcribed Image Text:

In this problem we will use a neural network to classify the crosses (x) from the circles (o) in the simple dataset shown in Figure [1(a). Even though the crosses and circles are not linearly separable, we can break the examples into three groups, S1, S2, and S3 (shown in Figure [1(a)) so that S is linearly separable from S and S is linearly separable from S3. We will exploit this fact to design weights for the neural network shown in Figure [1(b) in order to correctly classify this training set. For all nodes, we will use the threshold activation function x2 5, 4 3 x2 2 1 5 3 2 S2 S Wo x1 2 3 $(z) 3 x1 (a) The dataset with groups S1, S2, and S3. 4 (a) Set S2 and S3 = 41 1, z>0 0, 3 x2 2 z In this problem we will use a neural network to classify the crosses (x) from the circles (o) in the simple dataset shown in Figure [1(a). Even though the crosses and circles are not linearly separable, we can break the examples into three groups, S1, S2, and S3 (shown in Figure [1(a)) so that S is linearly separable from S and S is linearly separable from S3. We will exploit this fact to design weights for the neural network shown in Figure [1(b) in order to correctly classify this training set. For all nodes, we will use the threshold activation function x2 5, 4 3 x2 2 1 5 3 2 S2 S Wo x1 2 3 $(z) 3 x1 (a) The dataset with groups S1, S2, and S3. 4 (a) Set S2 and S3 = 41 1, z>0 0, 3 x2 2 z