wayground logo

Free Printable Worksheets

Font size

S
M
L
XL
Worksheets

Machine Learning Multiple Choice Questions

Total questions: 98

Worksheet time: 49mins

Name
Class
Date
1.

If a neuron gets input [1,1] and weights [0.5,0.5], bias 0.5, has ReLU activation, output is?

a)

0.5

b)

0

c)

1

d)

0.5

2.

Behavior of a learning process is usually, with high loads of noise?

a)

-1.5

b)

1.5

c)

-0.5

d)

0.5

3.

A neuron has weights [0.3, 0.7, 0.1] and input [1, 2, 1], what is the net input value?

a)

2

b)

3

c)

4

d)

1

4.

If any attribute has 3 possible values, it needs how many bits to encode for classification?

a)

2

b)

3

c)

1

d)

4

5.

Supervised version splits must be?

a)

General and random

b)

Specific and useful

c)

Consistent with splits

d)

Trained using standard splits

6.

What is the output of arithmetic mean(3,5,7,11)?

a)

10

b)

14

c)

3

d)

6.5

7.

What is the result of standard deviation(2,1,1)?

a)

1

b)

2

c)

3

d)

0

8.

For input [1, 2] and weights [0.4, 0.6], compute weighted sum.

a)

0.4

b)

0.6

c)

1.6

d)

0

9.

What is the output of sigmoid(1)?

a)

0.73

b)

0.5

c)

0.3

d)

1

10.

What is the output of ReLU([-1, 0, 1, 2])?

a)

[0, 0, 1, 2]

b)

[1, 0, 1, 2]

c)

[1, 1, 1, 2]

d)

[0, 1, 1, 2]

11.

If model output is not increasing, possible reason is?

a)

Convergence

b)

Divergence

c)

Overfitting

d)

Error increase

12.

Ensemble learning Stumps and original classifier both fail, reason might be?

a)

Exploding gradients

b)

Overfitting

c)

Structural incoherence

d)

No improvement

13.

Random forest improves accuracy for large data sets, reason is?

a)

Slaves to majority

b)

Overfitting

c)

Precision increases

d)

Noise-free data

14.

Bias & variance is optimal under which condition?

a)

Feature independence

b)

Large data

c)

High dimension

d)

Noise-free data

15.

Which difference between 10D/20D is significant?

a)

Bias

b)

Variance

c)

Noise

d)

Error

16.

Which algorithm defines boundaries between general and specific hypotheses?

a)

Candidate Elimination

b)

Decision Tree

c)

Naive Bayes

d)

SVM

17.

Which learning approach uses Naive Bayes classifier?

a)

Reinforcement learning

b)

Naive Bayes

c)

KNN

d)

SVM

18.

What is a 'hypothesis' in case-based learning?

a)

A function

b)

A problem

c)

A decision tree

d)

A case

19.

The goal of concept learning is to?

a)

Memorize data

b)

Boolean function

c)

Generalize

d)

Optimize variance

20.

What is the range of output in sigmoid activation?

a)

[0,1]

b)

[0,100]

c)

[0,20]

d)

[20,50]

21.

What is the output of softmax([10,20,30])?

a)

[0.03,0.04,0.93]

b)

[0.20,0.30,0.50]

c)

[0.10,0.20,0.30]

d)

[0.50,0.30,0.20]

22.

Which of the following is a problem for low bias models?

a)

High probability

b)

Overfitting

c)

Low variance

d)

Structural

23.

What type of model uncertainty can be handled using the 'mean ensemble'?

a)

Free flow

b)

Probability

c)

Structural

d)

Low variance

24.

In feature perception, model follows?

a)

Late computation

b)

Complex feature extraction

c)

Faster switching

d)

Thought-based

25.

Which metric measures angular separation?

a)

Euclidean

b)

Cosine

c)

Manhattan

d)

Chebyshev

26.

What is normalization in ML?

a)

100

b)

0

c)

1

d)

10

27.

A network, sigmoid node with output 0.5, what is the activation?

a)

0.5

b)

0

c)

1

d)

2

28.

Combination of odd/even with symmetry breaking is?

a)

Symmetry breaking

b)

Plans, outputs, losses

c)

Feature extraction

d)

Reproduction

29.

What is the final output in a genetic algorithm?

a)

Evaluation

b)

Mutation

c)

Reproduction

d)

Losses

30.

What is the second most used loss in ML apart from MSE?

a)

Size + location

b)

L1 loss

c)

L2 loss

d)

Data gradients

31.

Population = 50 and data = 10410^4 , how many test individuals are needed?

a)

10

b)

50

c)

100

d)

10310^3

32.

Model has similarity score=0.2 and 0.5 on two sets, what is the absolute difference in similarity?

a)

0.2

b)

0.3

c)

0.5

d)

0.7

33.

Bias affects which approximation in ML?

a)

Genetic algorithms

b)

Enough neurons

c)

Adaptive learning

d)

Data is linear

34.

Which operation keeps all keys from both join tables?

a)

Outer join

b)

Left join

c)

Right join

d)

Full join

35.

Which layer passes all except input in a neural network?

a)

Input layer

b)

Hidden layer

c)

Output layer

d)

All except input

36.

Why is test variance not matching train variance, reason may be?

a)

High test accuracy

b)

Overfitting

c)

Underfitting

d)

Optimal range

37.

What happens if one cannot decide best performance on test set?

a)

Tie

b)

Cannot decide

c)

Optimal

d)

Best

38.

For input [1,2] and weights [0.5,1], weighted sum is?

a)

2

b)

1.5

c)

1

d)

2.5

39.

In early ML why CNN training was difficult?

a)

No training

b)

No memory

c)

Online

d)

Uses genetic algorithms

40.

Training time doubles when data size increases from 1000 to 2000, if real time use, what is?

a)

16

b)

256

c)

16

d)

256

41.

BEE, edge from A to B implies:

a)

causes B

b)

is independent

c)

depends if independent

d)

No effect

42.

Which of the following is NOT a form of machine learning?

a)

Supervised

b)

Unsupervised

c)

Reinforced

d)

Associated

43.

If you randomly select a restaurant and try a dish, what type of learning is this?

a)

Return empty survey

b)

Create new file

c)

Return home

d)

Reinforced

44.

What is the main rule behind machine learning?

a)

Rule learning

b)

Instance correlation

c)

Weight correlation

d)

0-1

45.

For a model, if x=1, weights D=4, Q=4 and bias B=2, what is the linear model base equation?

a)

4x+2

b)

4x+4

c)

2x+4

d)

4x+0

46.

What does backpropagation adjust?

a)

Training data

b)

Number of layers

c)

Weights

d)

Inputs

47.

What will replace 0.5 in [0.25,0.5,0.75,1]?

a)

0.25

b)

0.33,1.34

c)

0.5,0.8,1

d)

0.6,0.8,1

48.

The gradient in most optimization functions is found as -0.1, -0.2, -0.3. What is the sequence?

a)

Error

b)

A

c)

B

d)

C

49.

If a data point is poorly chosen, what is the impact to accuracy?

a)

Improved accuracy

b)

Linear separation

c)

Failure to capture generalization

d)

Robust generalization

50.

Sample complexity is primarily important in:

a)

Hypothesis learning rate

b)

Test set

c)

Batch size

d)

A

51.

Match variable in case-based learning:

a)

Feature

b)

Feature dataset

c)

Match vectors

d)

B

52.

If threshold is 0.4 and model has 4 parents, the 5th parent, which is a spammer, is:

a)

Equal

b)

Not equal

c)

Depends on data

d)

A

53.

In perceptron, if set is w=0.6, 0.7, 0.8, 0.9, and average is 0.8, then:

a)

0.8

b)

0.6

c)

0.9

d)

A

54.

If each model process (p1, p11, p111, p1111) has equal test result with highest/lowest score, what is the output?

a)

k-Means

b)

Naive Bayes

c)

SVM

d)

B

55.

The naive hypothesis on probabilities: 0.1, 0.2, 0.3, 0.4. What is the expected probability?

a)

0.3

b)

0.15

c)

0.5

d)

C

56.

Find the principal axis, if p=0.5, q=0.5, find the projection.

a)

0.5

b)

0.5

c)

0.5

d)

C

57.

The principal function used for deep learning:

a)

Activation

b)

Learning rate

c)

Bias

d)

A

58.

For data set with missing values, frequent fill, what is the most common fill?

a)

Class-Row Mode

b)

Uniformity loss

c)

Consistency loss

d)

B

59.

If vector [4,2,1] performs what operation?

a)

Vector add

b)

Scalar loop

c)

Loop add

d)

Recursive add

60.

D=[2,4,6,8], what is D[2]?

a)

A) 4

b)

B) 6

c)

C) 2

d)

D) C

61.

D=[parentFn(1,2,1)+3,4,5], what is D[1]?

a)

4

b)

5

c)

3

d)

C

62.

D=[parentFn(1,2,1)+3,4,5], what is D[2]?

a)

5

b)

4

c)

3

d)

C

63.

D=[parentFn(1,2,1)+3,4,5], what is D[0]?

a)

5

b)

4

c)

6

d)

C

64.

4x1000+2x1000+1x1001 in octal representation is:

a)

10001

b)

11010

c)

10011

d)

C

65.

Which sampling method combines all points?

a)

pointQ

b)

pointP

c)

pointJ

d)

pointJ and pointA

66.

When using Gini index for splitting, which value gives maximum split?

a)

Minimum Gini index

b)

Maximum Gini index

c)

Equal split count

d)

Highest information

67.

When is the value of fake outcome clear?

a)

After network

b)

After network successful

c)

Anytime

d)

After network failed

68.

Which of the following correctly represents the average of the group (3,3,3)?

a)

mean(3,3,3)

b)

mean(3,3,3,1)

c)

mean(3,3,3,0)

d)

mean(3,3,3,4)

69.

What is the effect of choosing less training time?

a)

Less training time

b)

Simpler pattern consumption

c)

Better time fitting

d)

Problem restriction

70.

Adaptation phase in CBIR refers to:

a)

Image match

b)

Minor pattern selection

c)

Better time fitting

d)

Problem restriction

71.

What does a shape return for data frame with 3 rows and 4 columns?

a)

3,4

b)

4,3

c)

3,1

d)

1,3

72.

In genetic programming, individual trees represent:

a)

Vectors

b)

Numbers

c)

Layers

d)

A

73.

A hypothesis that covers all positive and negative instances is:

a)

Overfitting

b)

Underfitting

c)

Incomplete

d)

Consistent

74.

For a mean square of size 25, how many multiplications are there?

a)

25

b)

0

c)

5

d)

0

75.

What causes the memory of RBF to appear unstable?

a)

Uneven adaptation

b)

Adaptive weights

c)

All of the above

d)

Receptive field bias

76.

If classes are formed only on class names, but not on class values, what kind of induced risk behavior arises?

a)

Sigmoid

b)

ReLU

c)

Linear

d)

C

77.

Which exploration budget method for random agent application has highest reach in decision space?

a)

Random exploration

b)

Exhaustive checking

c)

Guided decision checking

d)

0.09

78.

Principal component analysis reduces dimensionality by:

a)

Reduced dimension

b)

Smaller

c)

Better for breaking

d)

0.09

79.

The total training data by 30% reduction, induced risk value is:

a)

Smaller

b)

Better for breaking

c)

Features are correlated

d)

0.09

80.

Mahalanobis distance is most useful for:

a)

Categorical

b)

Feature variance

c)

Features are correlated

d)

Feature are uncorrelated

81.

Ideal PCA has:

a)

Feature variance

b)

Feature are uncorrelated

c)

Features are correlated

d)

Feature are uncorrelated

82.

With a hypothesis H0(p), H1(p), p0 = 0.4, p1 = 0.6, what is the optimal expected probability of error?

a)

0.76

b)

0.68

c)

0.72

d)

0.84

83.

A hypothesis space of size 64, what is the number of bits required to encode the most consistent hypothesis in the space?

a)

44

b)

32

c)

8

d)

0

84.

Which is the most specific hypothesis in the hypothesis space?

a)

(0,0,0)

b)

(1,1,1)

c)

(0,X,0)

d)

(X,X,X)

85.

Which algorithm does not make a margin?

a)

Majority vote

b)

Random subspace

c)

SVM margin

d)

Bagging

86.

A hypothesis space H has 150 elements, what is the VC dimension of H?

a)

150

b)

149

c)

148

d)

147

87.

Which feature is not an example of a feature for a feedforward neural model?

a)

Distance measurement

b)

Label frequency

c)

Learning rate

d)

Probabilistic model

88.

Which feature cannot determine probabilistic model?

a)

Distance measurement

b)

Label frequency

c)

Learning rate

d)

Probabilistic model

89.

Which normalization function normalizes the sum to 1?

a)

Uses softmax

b)

Applies ReLU

c)

Units max

d)

Units min

90.

Which activation function is valid for a regression model?

a)

Linear

b)

Sigmoid

c)

ReLU

d)

Softmax

91.

A decision tree with 10 leaves, 2 classes, 2048 rows, what is the number of distinct root instances?

a)

2048

b)

512

c)

204

d)

4096

92.

Which function is used to balance data from imbalanced classes?

a)

Calculate node

b)

Select node

c)

Evaluate training matrix

d)

Choose split attribute

93.

A Bayesian belief model uses which tree?

a)

Tree

b)

Graph

c)

Matrix

d)

Table

94.

Which evaluation metric is not a precision metric?

a)

0.7

b)

0.9

c)

0.1

d)

0.13

95.

Which distance function is not a metric?

a)

Mahalanobis

b)

Cosine

c)

Euclidean

d)

Hamming

96.

A decision tree with 10 leaves, 2 classes, 2048 rows, what is the number of distinct root instances?

a)

2048

b)

512

c)

204

d)

4096

97.

Which function is used to balance data from imbalanced classes?

a)

Calculate node

b)

Select node

c)

Evaluate training matrix

d)

Choose split attribute

98.

Which evaluation metric is not a precision metric?

a)

0.7

b)

0.9

c)

0.1

d)

0.13