Data Mining | Week 6

Data Mining Week 6 Answers

link: Data Mining Week (nptel.ac.in)

Q1. Sufficient Number of output nodes required in an ANN used for two-class classification problem is:

(A) Random number
(B) Same as number of input nodes
(C) 1
(D) 2

Q2. How are the weights and biases initialized in an ANN in general?

(A) Can be initialized randomly
(B) Always initialized to zero
(C) Always initialized to infinity
(D) Always initialized as 1

Q3. In which of the below neural network, the links may connect nodes within the same layer or nodes from one layer to the previous layers?

(A) Perceptron
(B) Feed-forward neural network
(C) Recurrent neural network
(D) Both B, C

Q4. Neural Networks are complex ______________ with many parameters

(A) Linear Functions
(B) Nonlinear Functions
(C) Discrete Functions
(D) Exponential Functions

Q5. Artificial neural network used for:

(A) Pattern Recognition
(B) Classification
(C) Clustering
(D) All of the above

Q6. A neuron with 3 inputs has the weight vector [0.2 -0.1 0.1]^T and a bias θ = 0. If the input vector is X = [0.2 0.4 0.2]^T then the total input to the neuron is:

(A) 0.2
(B) 0.02
(C) 0.4
(D) 0.10

Q7. We are designing a SVM, W^Tx+b=0, suppose X/s are the support vectors and alpha / s the corresponding Lagrange multipliers, then which of the following statements are correct:

(A) AND
(B) NAND
(C) XOR
(D) NOR

Q8. The neural network given bellow takes two binary valued inputs x₁, x₂ ϵ {0,1}, the activation function for each neuron is the binary threshold function (g(a)= 1 if a >0; 0 otherwise). Which of the following logical functions does it compute?

(A) AND
(B) NAND
(C) XOR
(D) NOR

Q9. The neural network given bellow takes two binary valued inputs x₁, x₂ ϵ {0,1} and the activation function is the binary threshold function (h(z)=1 if z>0;0 otherwise). Which of the following logical functions does it compute?

(A) OR
(B) AND
(C) NAND
(D) NOR

Q10. Under which of the following situation would you expect overfitting to happen?

(A) With training iterations error on training set as well as test set decreases
(B) With training iterations error on training set decreases but the test set increases
(C) With training iterations error on training set as well as test set increases
(D) With training iterations training set as well as test set error remains constant