NEURAL NETWORK ASSIGNMENT
- Assume you want to classify the following four points (x1; x2) €IR2:
|
X1 |
0 |
0 |
1 |
1 |
|
X2 |
0 |
1 |
0 |
1 |
|
Class |
C0 |
C0 |
C0 |
C1 |
- Explain the simplest neural network that can classify the above points?
- Consider the following points
Note: Fully specify all necessary parameters of a neural network that can classify the points in Part (a) and draw the decision boundary.(6marks)
|
X1 |
0 |
0 |
1 |
1 |
|
X2 |
0 |
1 |
0 |
1 |
|
Class |
C0 |
C1 |
C1 |
C0 |
Explain whether the network you provided in part (a) correctly classify these points. (4marks)
2. Back-propagation algorithm:
Consider the following neural network:
With configuration:-
|
1 |
1 |
1 |
1 |
1 |
-1 |
0.5 |
-0.5 |
-0.5 |
Perform one iteration of the back propagation algorithm, assuming that the hidden layer and the output layer use the Log-Sigmoid activation function.
yk = —————————————————(1)
for inputs (x1; x2) = (0,1) and output y = 1. (6marks)
3. Prove that if the activation function of a 2 layered neural network is the identity function, the neural network is equivalent to a simple perceptron with a linear activation function.(4marks)
0 comments